1.6.1. Introduction

When a binary linked against the library is executed, the Cygwin DLL is loaded into the application's text segment. Because we are trying to emulate a UNIX kernel which needs access to all processes running under it, the first Cygwin DLL to run creates shared memory areas that other processes using separate instances of the DLL can access. This is used to keep track of open file descriptors and assist fork and exec, among other purposes. In addition to the shared memory regions, every process also has a per_process structure that contains information such as process id, user id, signal masks, and other similar process-specific information.

The DLL is implemented using the Win32 API, which allows it to run on all Win32 hosts. Because processes run under the standard Win32 subsystem, they can access both the UNIX compatibility calls provided by Cygwin as well as any of the Win32 API calls. This gives the programmer complete flexibility in designing the structure of their program in terms of the APIs used. For example, they could write a Win32-specific GUI using Win32 API calls on top of a UNIX back-end that uses Cygwin.

Early on in the development process, we made the important design decision that it would not be necessary to strictly adhere to existing UNIX standards like POSIX.1 if it was not possible or if it would significantly diminish the usability of the tools on the Win32 platform. In many cases, an environment variable can be set to override the default behavior and force standards compliance.

1.6.2. Supporting both Windows NT and 9x

While Windows 95 and Windows 98 are similar enough to each other that we can safely ignore the distinction when implementing Cygwin, Windows NT is an extremely different operating system. For this reason, whenever the DLL is loaded, the library checks which operating system is active so that it can act accordingly.

In some cases, the Win32 API is only different for historical reasons. In this situation, the same basic functionality is available under Windows 9x and NT but the method used to gain this functionality differs. A trivial example: in our implementation of uname, the library examines the sysinfo.dwProcessorType structure member to figure out the processor type under Windows 9x. This field is not supported in NT, which has its own operating system-specific structure member called sysinfo.wProcessorLevel.

Other differences between NT and 9x are much more fundamental in nature. The best example is that only NT provides a security model.

1.6.3. Permissions and Security

Windows NT includes a sophisticated security model based on Access Control Lists (ACLs). Cygwin maps Win32 file ownership and permissions to the more standard, older UNIX model by default. Cygwin version 1.1 introduces support for ACLs according to the system calls used on newer versions of Solaris. This ability is used when the `ntsec' feature is switched on which is described in another chapter. The chmod call maps UNIX-style permissions back to the Win32 equivalents. Because many programs expect to be able to find the /etc/passwd and /etc/group files, we provide utilities that can be used to construct them from the user and group information provided by the operating system.

Under Windows NT, the administrator is permitted to chown files. There is no mechanism to support the setuid concept or API call since Cygwin version 1.1.2. With version 1.1.3 Cygwin introduces a mechanism for setting real and effective UIDs under Windows NT/W2K. This is described in the ntsec section.

Under Windows 9x, the situation is considerably different. Since a security model is not provided, Cygwin fakes file ownership by making all files look like they are owned by a default user and group id. As under NT, file permissions can still be determined by examining their read/write/execute status. Rather than return an unimplemented error, under Windows 9x, the chown call succeeds immediately without actually performing any action whatsoever. This is appropriate since essentially all users jointly own the files when no concept of file ownership exists.

It is important that we discuss the implications of our "kernel" using shared memory areas to store information about Cygwin processes. Because these areas are not yet protected in any way, in principle a malicious user could modify them to cause unexpected behavior in Cygwin processes. While this is not a new problem under Windows 9x (because of the lack of operating system security), it does constitute a security hole under Windows NT. This is because one user could affect the Cygwin programs run by another user by changing the shared memory information in ways that they could not in a more typical WinNT program. For this reason, it is not appropriate to use Cygwin in high-security applications. In practice, this will not be a major problem for most uses of the library.

1.6.4. File Access

Cygwin supports both Win32- and POSIX-style paths, using either forward or back slashes as the directory delimiter. Paths coming into the DLL are translated from Win32 to POSIX as needed. As a result, the library believes that the file system is a POSIX-compliant one, translating paths back to Win32 paths whenever it calls a Win32 API function. UNC pathnames (starting with two slashes) are supported.

The layout of this POSIX view of the Windows file system space is stored in the Windows registry. While the slash ('/') directory points to the system partition by default, this is easy to change with the Cygwin mount utility. In addition to selecting the slash partition, it allows mounting arbitrary Win32 paths into the POSIX file system space. Many people use the utility to mount each drive letter under the slash partition (e.g. C:\ to /c, D:\ to /d, etc...).

The library exports several Cygwin-specific functions that can be used by external programs to convert a path or path list from Win32 to POSIX or vice versa. Shell scripts and Makefiles cannot call these functions directly. Instead, they can do the same path translations by executing the cygpath utility program that we provide with Cygwin.

Win32 file systems are case preserving but case insensitive. Cygwin does not currently support case distinction because, in practice, few UNIX programs actually rely on it. While we could mangle file names to support case distinction, this would add unnecessary overhead to the library and make it more difficult for non-Cygwin applications to access those files.

Symbolic links are emulated by files containing a magic cookie followed by the path to which the link points. They are marked with the System attribute so that only files with that attribute have to be read to determine whether or not the file is a symbolic link. Hard links are fully supported under Windows NT on NTFS file systems. On a FAT file system, the call falls back to simply copying the file, a strategy that works in many cases.

The inode number for a file is calculated by hashing its full Win32 path. The inode number generated by the stat call always matches the one returned in d_ino of the dirent structure. It is worth noting that the number produced by this method is not guaranteed to be unique. However, we have not found this to be a significant problem because of the low probability of generating a duplicate inode number.

Chroot is supported since release 1.1.3. Note that chroot isn't supported native by Windows. This implies some restrictions. First of all, the chroot call isn't a privileged call. Each user may call it. Second, the chroot environment isn't safe against native windows processes. If you want to support a chroot environment as, for example, by allowing an anonymous ftp with restricted access, you'll have to care that only native Cygwin applications are accessible inside of the chroot environment. Since that applications are only using the Cygwin POSIX API to access the file system their access can be restricted as it is intended. This includes not only POSIX paths but Win32 paths (containing drive letter and/or backslashes) and CIFS paths (//server/share or \\server\share) as well.

1.6.5. Text Mode vs. Binary Mode

Interoperability with other Win32 programs such as text editors was critical to the success of the port of the development tools. Most Red Hat customers upgrading from the older DOS-hosted toolchains expected the new Win32-hosted ones to continue to work with their old development sources.

Unfortunately, UNIX and Win32 use different end-of-line terminators in text files. Consequently, carriage-return newlines have to be translated on the fly by Cygwin into a single newline when reading in text mode.

This solution addresses the compatibility requirement at the expense of violating the POSIX standard that states that text and binary mode will be identical. Consequently, processes that attempt to lseek through text files can no longer rely on the number of bytes read as an accurate indicator of position in the file. For this reason, the CYGWIN environment variable can be set to override this behavior.

1.6.6. ANSI C Library

We chose to include Red Hat's own existing ANSI C library "newlib" as part of the library, rather than write all of the lib C and math calls from scratch. Newlib is a BSD-derived ANSI C library, previously only used by cross-compilers for embedded systems development.

The reuse of existing free implementations of such things as the glob, regexp, and getopt libraries saved us considerable effort. In addition, Cygwin uses Doug Lea's free malloc implementation that successfully balances speed and compactness. The library accesses the malloc calls via an exported function pointer. This makes it possible for a Cygwin process to provide its own malloc if it so desires.

1.6.7. Process Creation

The fork call in Cygwin is particularly interesting because it does not map well on top of the Win32 API. This makes it very difficult to implement correctly. Currently, the Cygwin fork is a non-copy-on-write implementation similar to what was present in early flavors of UNIX.

The first thing that happens when a parent process forks a child process is that the parent initializes a space in the Cygwin process table for the child. It then creates a suspended child process using the Win32 CreateProcess call. Next, the parent process calls setjmp to save its own context and sets a pointer to this in a Cygwin shared memory area (shared among all Cygwin tasks). It then fills in the child's .data and .bss sections by copying from its own address space into the suspended child's address space. After the child's address space is initialized, the child is run while the parent waits on a mutex. The child discovers it has been forked and longjumps using the saved jump buffer. The child then sets the mutex the parent is waiting on and blocks on another mutex. This is the signal for the parent to copy its stack and heap into the child, after which it releases the mutex the child is waiting on and returns from the fork call. Finally, the child wakes from blocking on the last mutex, recreates any memory-mapped areas passed to it via the shared area, and returns from fork itself.

While we have some ideas as to how to speed up our fork implementation by reducing the number of context switches between the parent and child process, fork will almost certainly always be inefficient under Win32. Fortunately, in most circumstances the spawn family of calls provided by Cygwin can be substituted for a fork/exec pair with only a little effort. These calls map cleanly on top of the Win32 API. As a result, they are much more efficient. Changing the compiler's driver program to call spawn instead of fork was a trivial change and increased compilation speeds by twenty to thirty percent in our tests.

However, spawn and exec present their own set of difficulties. Because there is no way to do an actual exec under Win32, Cygwin has to invent its own Process IDs (PIDs). As a result, when a process performs multiple exec calls, there will be multiple Windows PIDs associated with a single Cygwin PID. In some cases, stubs of each of these Win32 processes may linger, waiting for their exec'd Cygwin process to exit.

1.6.8. Signals

When a Cygwin process starts, the library starts a secondary thread for use in signal handling. This thread waits for Windows events used to pass signals to the process. When a process notices it has a signal, it scans its signal bitmask and handles the signal in the appropriate fashion.

Several complications in the implementation arise from the fact that the signal handler operates in the same address space as the executing program. The immediate consequence is that Cygwin system functions are interruptible unless special care is taken to avoid this. We go to some lengths to prevent the sig_send function that sends signals from being interrupted. In the case of a process sending a signal to another process, we place a mutex around sig_send such that sig_send will not be interrupted until it has completely finished sending the signal.

In the case of a process sending itself a signal, we use a separate semaphore/event pair instead of the mutex. sig_send starts by resetting the event and incrementing the semaphore that flags the signal handler to process the signal. After the signal is processed, the signal handler signals the event that it is done. This process keeps intraprocess signals synchronous, as required by POSIX.

Most standard UNIX signals are provided. Job control works as expected in shells that support it.

1.6.9. Sockets

Socket-related calls in Cygwin simply call the functions by the same name in Winsock, Microsoft's implementation of Berkeley sockets. Only a few changes were needed to match the expected UNIX semantics - one of the most troublesome differences was that Winsock must be initialized before the first socket function is called. As a result, Cygwin has to perform this initialization when appropriate. In order to support sockets across fork calls, child processes initialize Winsock if any inherited file descriptor is a socket.

Unfortunately, implicitly loading DLLs at process startup is usually a slow affair. Because many processes do not use sockets, Cygwin explicitly loads the Winsock DLL the first time it calls the Winsock initialization routine. This single change sped up GNU configure times by thirty percent.

1.6.10. Select

The UNIX select function is another call that does not map cleanly on top of the Win32 API. Much to our dismay, we discovered that the Win32 select in Winsock only worked on socket handles. Our implementation allows select to function normally when given different types of file descriptors (sockets, pipes, handles, and a custom /dev/windows Windows messages pseudo-device).

Upon entry into the select function, the first operation is to sort the file descriptors into the different types. There are then two cases to consider. The simple case is when at least one file descriptor is a type that is always known to be ready (such as a disk file). In that case, select returns immediately as soon as it has polled each of the other types to see if they are ready. The more complex case involves waiting for socket or pipe file descriptors to be ready. This is accomplished by the main thread suspending itself, after starting one thread for each type of file descriptor present. Each thread polls the file descriptors of its respective type with the appropriate Win32 API call. As soon as a thread identifies a ready descriptor, that thread signals the main thread to wake up. This case is now the same as the first one since we know at least one descriptor is ready. So select returns, after polling all of the file descriptors one last time.

2.1.1. Download Source

Cygwin uses packages to manage installing various software. When the default Install from Internet option is chosen, setup.exe creates a local directory to store the packages before actually installing the contents. Download from Internet performs only the first part (storing the packages locally), while Install from Local Directory performs only the second (installing the contents of the packages).

The Download from Internet option is mainly for creating a base Cygwin package tree on one computer for installation on several machines with Install from Local Directory; copy the entire local package tree to another machine with the directory tree intact. For example, you might create a C:\cache\ directory and place setup.exe in it. Run setup.exe to Install from Internet or Download from Internet, then copy the whole C:\cache\ to each machine and instead choose Install from Local Directory. Unfortunately setup.exe does not yet support unattended installs.

Though this provides some basic mirroring functionality, if you are managing a wide Cygwin installation, to keep up to date we recommend using a mirroring tool such as wget. A helpful user on the Cygwin mailing list created a simple demonstration script to accomplish this; search the list for mkcygwget for ideas.

2.1.2. Selecting an Install Directory

The Root Directory for Cygwin (default C:\cygwin) will become / within your Cygwin installation. You must have write access to the parent directory, and any ACLs on the parent directory will determine access to installed files.

The Install For options of All Users or Just Me should always be left on the default All Users, unless you do not have write access to HKEY_LOCAL_MACHINE in the registry or the All Users Start Menu. This is true even if you are the only user planning to use Cygwin on the machine. Selecting Just Me will cause problems for programs such as crond and sshd. If you do not have the necessary permissions, but still want to use these programs, consult the Cygwin mailing list archives about others' experiences.

The Default Text File Type should be left on Unix (that is, \n) unless you have a very good reason to switch it to DOS (that is, \r\n).

2.1.3. Local Package Directory

The Local Package Directory is the cache where setup.exe stores the packages before they are installed. The cache must not be the same folder as the Cygwin root. Within the cache, a separate directory is created for each Cygwin mirror, which allows setup.exe to use multiple mirrors and custom packages. After installing Cygwin, the cache is no longer necessary, but you may want to retain the packages as backups, for installing Cygwin to another system, or in case you need to reinstall a package.

2.1.4. Connection Method

The Direct Connection method of downloading will directly download the packages, while the IE5 method will leverage your IE5 cache for performance. If your organisation uses a proxy server or auto-configuration scripts, the IE5 method also uses these settings. If you have a proxy server, you can manually type it into the Use Proxy section. Unfortunately, setup.exe does not currently support password authorization for proxy servers.

2.1.5. Choosing Mirrors

Since there is no way of knowing from where you will be downloading Cygwin, you need to choose at least one mirror site. Cygwin mirrors are geographically distributed around the world; check the list at http://cygwin.com/mirrors.html to find one near you. You can select multiple mirrors by holding down CTRL and clicking on each one. If you have the URL of an unlisted mirror (for example, if your organization has an internal Cygwin mirror) you can add it.

2.1.6. Choosing Packages

For each selected mirror site, setup.exe downloads a small text file called setup.bz2 that contains a list of packages available from that site along with some basic information about each package which setup.exe parses and uses to create the chooser window. For details about the format of this file, see http://sources.redhat.com/cygwin-apps/setup.html.

The chooser is the most complex part of setup.exe. Packages are grouped into categories, and one package may belong to multiple categories (assigned by the volunteer package maintainer). Each package can be found under any of those categories in the heirarchial chooser view. By default setup.exe will install only the packages in the Base category and their dependencies, resulting in a minimal Cygwin installation. However, this will not include many commonly used tools such as gcc (which you will find in the Devel category).

You can change setup.exe's view style, which is helpful if you know the name of a package you want to install but not which category it is in. Click on the View button and it will rotate between Category (the default), Full (all packages), and Partial (only packages to be upgraded). If you are familiar with Unix, you will probably want to at least glance through the Full listing for your favorite tools.

Once you have an existing Cygwin installation, the setup.exe chooser is also used to manage your Cygwin installation. Information on installed packages is kept in the /etc/setup/ directory of your Cygwin installation; if setup.exe cannot find this directory it will act just like you had no Cygwin installation. If setup.exe finds a newer version of an installed package available, it will automatically mark it to be upgraded. To Uninstall, Reinstall, or get the Source for an existing package, click on Keep to toggle it. Also, to avoid the need to reboot after upgrading, make sure to close all Cygwin windows and stop all Cygwin processes before setup.exe begins to install the upgraded package.

The final feature of the setup.exe chooser is for Previous and Experimental packages. By default the chooser shows only the current version of each package, though mirrors have at least one previous version and occasionally there is a testing or beta version of a package available. To see these package, click on the Prev or Exp radio button. Be warned, however, that the next time you run setup.exe it will try to replace old or experimental versions with the latest.

2.1.7. Download and Installation Progress

First, setup.exe will download all selected packages to the local directory chosen earlier. Before installing, setup.exe performs a checksum on each package. If the local directory is a slow medium (such as a network drive) this can take a long time. During the download and installation, setup.exe show progress bars for the current task and total remaining disk space.

2.1.8. Icons

You may choose to install shortcuts on the Desktop and/or Start Menu to start a bash shell. If you prefer to use a different shell or the native Windows version of rxvt, you can use these shortcuts as a guide to creating your own.

2.1.9. Post-Install Scripts

Last of all, setup.exe will run any post-install scripts to finish correctly setting up installed packages. Since each script is run separately, several windows may pop up. If you are interested in what is being done, see the Cygwin Package Contributor's Guide at http://cygwin.com/setup.html When the last post-install script is completed, setup.exe will display a box announcing the completion. A few packages, such as the OpenSSH server, require some manual site-specific configuration. Relevant documentation can be found in the /usr/doc/Cygwin/ directory.

2.4.1. NT security

The NT security allows a process to allow or deny access of different kind to `objects'. `Objects' are files, processes, threads, semaphores, etc.

The main data structure of NT security is the `security descriptor' (SD) structure. It explains the permissions, that are granted (or denied) to an object and contains information, that is related to so called `security identifiers' (SID).

A SID is a unique identifier for users, groups and domains. SIDs are comparable to UNIX UIDs and GIDs, but are more complicated because they are unique across networks. Example:

SID of a system `foo':

  S-1-5-21-165875785-1005667432-441284377

SID of a user `johndoe' of the system `foo':

  S-1-5-21-165875785-1005667432-441284377-1023

The above example shows the convention for printing SIDs. The leading `S' should show that it is a SID. The next number is a version number which is always 1. The next number is the so called `top-level authority' that identifies the source that issued the SID.

While each system in a NT network has it's own SID, the situation is modified in NT domains: The SID of the domain controller is the base SID for each domain user. If an NT user has one account as domain user and another account on his local machine, this accounts are under any circumstances DIFFERENT, regardless of the usage of the same user name and password!

SID of a domain `bar':

  S-1-5-21-186985262-1144665072-740312968

SID of a user `johndoe' in the domain `bar':

  S-1-5-21-186985262-1144665072-740312968-1207

The last part of the SID, the so called `relative identifier' (RID), is by default used as UID and/or GID under cygwin. As the name and the above example implies, this id is unique only relative to one system or domain.

Note, that it's possible, that an user has the same RID on two different systems. The resulting SIDs are nevertheless different, so the SIDs are representing different users in an NT network.

There is a big difference between UNIX IDs and NT SIDs, the existence of the so called `well known groups'. For example UNIX has no GID for the group of `all users'. NT has an SID for them, called `Everyone' in the English versions. The SIDs of well-known groups are not unique across an NT network but their meanings are unmistakable. Examples of well-known groups:

everyone                        S-1-1-0
creator/owner                   S-1-3-0
batch process (via `at')        S-1-5-3
authenticated users             S-1-5-11
system                          S-1-5-18

The last important group of SIDs are the `predefined groups'. This groups are used mainly on systems outside of domains to simplify the administration of user permissions. The corresponding SIDs are not unique across the network so they are interpreted only locally:

administrators                  S-1-5-32-544
users                           S-1-5-32-545
guests                          S-1-5-32-546
...

Now, how are permissions given to objects? A process may assign an SD to the object. The SD of an object consists of three parts:

• the SID of the owner

• the SID of the group

• a list of SIDs with their permissions, called `access control list' (ACL)

UNIX is able to create three different permissions, the permissions for the owner, for the group and for the world. In contrast the ACL has a potentially infinite number of members. Every member is a so called `access control element' (ACE). An ACE contains three parts:

• the type of the ACE

• permissions, described with a DWORD

• the SID, for which the above mentioned permissions are set

The two important types of ACEs are the `access allowed ACE' and the `access denied ACE'. The ntsec functionality only used `access allowed ACEs' up to Cygwin version 1.1.0. Later versions also use `access denied ACEs' to reflect the UNIX permissions as well as possible.

The possible permissions on objects are more detailed than in UNIX. For example, the permission to delete an object is different from the write permission.

With the aforementioned method NT is able to grant or revoke permissions to objects in a far more specific way. But what about cygwin? In a POSIX environment it would be fine to have the security behavior of a POSIX system. The NT security model is MOSTLY able to reproduce the POSIX model. The ntsec method tries to do this in cygwin.

You ask "Mostly? Why mostly???" Because there's a leak in the NT model. I will describe that in detail in chapter 4.

Creating explicit object security is not that easy so you will often see only two simple variations in use:

• default permissions, computed by the operating system

• each permission to everyone

For parameters to functions that create or open securable objects another data structure is used, the `security attributes' (SA). This structure contains an SD and a flag that specifies whether the returned handle to the object is inherited to child processes or not. This property is not important for ntsec so in this document the difference between SDs and SAs is ignored.

2.4.2. Process privileges

Any process started under control of cygwin has a semaphore attached to it, that is used for signaling purposes. The creation of this semaphore can be found in sigproc.cc, function `getsem'. The first parameter to the function call `CreateSemaphore' is an SA. Without ntsec this SA assigns default security to the semaphore. There is a simple disadvantage: Only the owner of the process may send signals to it. Or, in other words, if the owner of the process is not a member of the administrators' group, no administrator may kill the process! This is especially annoying, if processes are started via service manager.

Ntsec now assigns an SA to the process control semaphore, that has each permission set for the user of the process, for the administrators' group and for `system', which is a synonym for the operating system itself. The creation of this SA is done by the function `sec_user', that can be found in `shared.cc'. Each member of the administrators' group is now allowed to send signals to any process created in cygwin, regardless of the process owner.

Moreover, each process now has the appropriate security settings, when it is started via `CreateProcess'. You will find this in function `spawn_guts' in module `spawn.cc'. The security settings for starting a process in another user context have to add the sid of the new user, too. In the case of the `CreateProcessAsUser' call, sec_user creates an SA with an additional entry for the sid of the new user.

2.4.3. File permissions

If ntsec is turned on, file permissions are set as in UNIX. An SD is assigned to the file containing the owner and group and ACEs for the owner, the group and `Everyone'.

The complete settings of UNIX like permissions can be found in the file `security.cc'. The two functions `get_nt_attribute' and `set_nt_attribute' are the main code. The reading and writing of the SDs is done by the functions `read_sd' and `write_sd'. `write_sd' uses the function `BackupRead' instead of the simpler function `SetFileSecurity' because the latter is unable to set owners different from the caller.

If you are creating a file `foo' outside of cygwin, you will see something like the following on ls -ln:

If your login is member of the administrators' group:

  rwxrwxrwx 1  544  513  ... foo

if not:

  rwxrwxrwx 1  1000  513  ... foo

Note the user and group IDs. 544 is the UID of the administrators' group. This is a `feature' :-P of WinNT. If one is a member of the administrators' group, every file, that he has created is owned by the administrators' group, instead by him.

The second example shows the UID of the first user, that has been created with NT's the user administration tool. The users and groups are sequentially numbered, starting with 1000. Users and groups are using the same numbering scheme, so a user and a group don't share the same ID.

In both examples the GID 513 is of special interest. This GID is a well known group with different naming in local systems and domains. Outside of domains the group is named 'None' (`Kein' in German, `Aucun' in French, etc.), in domains it is named 'Domain Users'. Unfortunately, the group `None' is never shown in the user admin tool outside of domains! This is very confusing but this seems to have no negative consequences.

To work correctly, ntsec depends on the files /etc/passwd/ and /etc/group. In cygwin release 1.0 the names and the IDs must correspond to the appropriate NT IDs! The IDs used in cygwin are the RID of the NT SID, as mentioned earlier. An SID of e.g. the user `corinna' on my NT workstation:

  S-1-5-21-165875785-1005667432-441284377-1000

Note the last number: It's the RID 1000, the cygwin's UID.

Unfortunately, workstations and servers outside of domains are not able to set primary groups! In these cases, where there is no correlation of users to primary groups, NT returns 513 (None) as primary group, regardless of the membership to existing local groups.

When using mkpasswd -l -g on such systems, you have to change the primary group by hand if `None' as primary group is not what you want (and I'm sure, it's not what you want!)

Look at the following examples, which were parts of my files before storing SIDs in /etc/passwd and /etc/group had been introduced (See next chapter for details). With the exception of my personal user entry, all entries are well known entries.

everyone:*:0:0:::
system:*:18:18:::
administrator::500:544::/home/root:/bin/bash
guest:*:501:546:::
administrators:*:544:544::/home/root:
corinna::1000:547:Corinna Vinschen:/home/corinna:/bin/tcsh

everyone::0:
system::18:
none::513:
administrators::544:
users::545:
guests::546:
powerusers::547:

As you can see, I changed my primary group membership from 513 (None) to 547 (powerusers). So all files I created inside of Cygwin were now owned by the powerusers group instead of None. This is the way I liked it.

Groups may be mentioned in the passwd file, too. This has two advantages:

• Because NT assigns them to files as owners, a ls -l is often more readable.

• Moreover it's possible to assigned them to files as owners with cygwin's chown.

The group `system' is the aforementioned synonym for the operating system itself and is normally the owner of processes, that are started through service manager. The same is true for files, that are created by processes, which are started through service manager.

2.4.4. New since Cygwin release 1.1

In Cygwin release 1.1 a new technique of using the /etc/passwd and /etc/group is introduced.

Both files may now contain SIDs of users and groups. They are saved in the last field of pw_gecos in /etc/passwd and in the gr_passwd field in /etc/group.

This has the following advantages:

• ntsec works better in domain environments.

• Accounts (users and groups) may get another name in cygwin than their NT account name. The name in /etc/passwd or /etc/group is transparently used by cygwin applications (eg. chown, chmod, ls):

  root::500:513::/home/root:/bin/sh

  instead of

  adminstrator::500:513::/home/root:/bin/sh

  Caution: If you like to use the account as login account via telnet etc. you have to remain the name unchanged or you have to use the special version of login which is part of the standard Cygwin distribution since 1.1.

• Cygwin UIDs and GIDs are now not necessarily the RID part of the NT SID:

  root::0:513:S-1-5-21-54355234-56236534-345635656-500:/home/root:/bin/sh

  instead of

  root::500:513::/home/root:/bin/sh

• As in U*X systems UIDs and GIDs numbering scheme now don't influence each other. So it's possible to have same Id's for a user and a group:

  Example 2-3. /etc/passwd:

  root::0:0:S-1-5-21-54355234-56236534-345635656-500:/home/root:/bin/sh

  Example 2-4. /etc/group:

  root:S-1-5-32-544:0:

The tools mkpasswd and mkgroup create the needed entries by default. If you don't want that you can use the options -s or --no-sids. I suggest not to do this since ntsec works better when having the SIDs available.

Please note that the pw_gecos field in /etc/passwd is defined as a comma seperated list. The SID has to be the last field!

As aforementioned you are able to use cygwin account names different from the NT account names. If you want to login thru `telnet' or something else you have to use the special login. You may then add another field to pw_gecos which contains the NT user name including it's domain. So you are able to login as each domain user. The syntax is easy: Just add an entry of the form U-ntdomain\ntusername to the pw_gecos field. Note that the SID must still remain the last field in pw_gecos!

the_king::1:1:Elvis Presley,U-STILLHERE\elvis,S-1-5-21-1234-5678-9012-1000:/bin/sh

For a local user just drop the domain:

the_king::1:1:Elvis Presley,U-elvis,S-1-5-21-1234-5678-9012-1000:/bin/sh

In either case the password of the user is taken from the NT user database, NOT from the passwd file!

As in the previous chapter I give my personal /etc/passwd and /etc/group as examples. Please note that I've changed these files heavily! There's no need to change them that way, it's just for testing purposes and... for fun.

root:*:0:0:Administrators group,S-1-5-32-544::
SYSTEM:*:18:18:,S-1-5-18:/home/system:/bin/bash
admin:*:500:513:,S-1-5-21-1844237615-436374069-1060284298-500:/home/Administrator:/bin/bash
corinna:*:100:0:Corinna Vinschen,S-1-5-21-1844237615-436374069-1060284298-1003:/home/corinna:/bin/tcsh
Guest:*:501:546:,S-1-5-21-1844237615-436374069-1060284298-501:/home/Guest:/bin/bash

root:S-1-5-32-544:0:
local:S-1-2-0:2:
network:S-1-5-2:3:
interactive:S-1-5-4:4:
authenticatedusers:S-1-5-11:5:
SYSTEM:S-1-5-18:18:
local_svc:S-1-5-19:19:
netwrk_svc:S-1-5-20:20:
none:S-1-5-21-1844237615-436374069-1060284298-513:513:
bckup_op:S-1-5-32-551:551:
guests:S-1-5-32-546:546:
pwrusers:S-1-5-32-547:547:
replicator:S-1-5-32-552:552:
users:S-1-5-32-545:545:

If you want to do similar changes to your files, please do that only if you're feeling comfortably with the concepts. Otherwise don't be surprised if some stuff doesn't work anymore. If you screwed up things, revert to files created by mkpasswd and mkgroup. Especially don't change the uid or the name of user SYSTEM. Even if that works mostly, some Cygwin applications running as local service under that account could behave strangly suddenly.

2.4.5. The mapping leak

Now its time to point out the leak in the NT permissions. The official documentation explains in short the following:

• access allow ACEs are accumulated regarding to the group membership of the caller.

• The order of ACEs is important. The system reads them in sequence until either any needed right is denied or all needed rights are granted. Later ACEs are then not taken into account.

• All access denied ACEs _should_ precede any access allowed ACE.

Note that the last rule is a preference, not a law. NT will correctly deal with the ACL regardless of the sequence order. The second rule is not modified to get the ACEs in the prefered order.

Unfortunately the security tab of the NT4 explorer is completely unable to deal with access denied ACEs while the explorer of W2K rearranges the order of the ACEs before you can read them. Thank God, the sort order remains unchanged if one presses the Cancel button.

You still ask "Where is the leak?" NT ACLs are unable to reflect each possible combination of POSIX permissions. Example:

rw-r-xrw-

1st try:

UserAllow:   110
GroupAllow:  101
OthersAllow: 110

Hmm, because of the accumulation of allow rights the user may execute because the group may execute.

2st try:

UserDeny:    001
GroupAllow:  101
OthersAllow: 110

Now the user may read and write but not execute. Better? No! Unfortunately the group may write now because others may write.

3rd try:

UserDeny:    001
GroupDeny:   010
GroupAllow:  001
OthersAllow: 110

Now the group may not write as intended but unfortunately the user may not write anymore, too. How should this problem be solved? According to the official rules a UserAllow has to follow the GroupDeny but it's easy to see that this can never be solved that way.

The only chance:

UserDeny:    001
UserAllow:   010
GroupDeny:   010
GroupAllow:  001
OthersAllow: 110

Again: This works for both, NT4 and W2K. Only the GUIs aren't able to deal with that order.

2.4.6. New acl API

For dealing with ACLs Cygwin now has the acl API as it's implemented in newer versions of Solaris. The new data structure for a single ACL entry (ACE in NT terminology) is defined in sys/acl.h as:

typedef struct acl {
  int     a_type;  /* entry type */
  uid_t   a_id;    /* UID | GID */
  mode_t  a_perm;  /* permissions */
} aclent_t;

The a_perm member of the aclent_t type contains only the bits for read, write and execute as in the file mode. If eg. read permission is granted, all read bits (S_IRUSR, S_IRGRP, S_IROTH) are set. CLASS_OBJ or MASK ACL entries are not fully implemented yet.

The new API calls are

acl(2), facl(2)
aclcheck(3),
aclsort(3),
acltomode(3), aclfrommode(3),
acltopbits(3), aclfrompbits(3),
acltotext(3), aclfromtext(3)

Like in Solaris, Cygwin has two new commands for working with ACLs on the command line: getfacl and setfacl.

Online man pages for the aforementioned commands and API calls can be found on eg. http://docs.sun.com

2.4.7. New setuid concept

UNIX applications which have to switch the user context are using the setuid and seteuid calls which are not part of the Windows API. Nevertheless these calls are supported under Windows NT/W2K since Cygwin release 1.1.3. Because of the nature of NT security an application which needs the ability has to be patched, though.

NT uses so called `access tokens' to identify a user and it's permissions. To switch the user context the application has to request such an `access token'. This is typically done by calling the NT API function LogonUser. The access token is returned and either used in ImpersonateLoggedOnUser to change user context of the current process or in CreateProcessAsUser to change user context of a spawned child process. An important restriction is that the application using LogonUser must have special permissions:

"Act as part of the operating system"
"Replace process level token"
"Increase quotas"

Note that administrators do not have all these user rights set by default.

Two new Cygwin calls are introduced to support porting setuid applications with a minimum of effort. You only give Cygwin the right access token and then you can call seteuid or setuid as usual in POSIX applications. The call to sexec is not needed anymore. Porting a setuid application is illustrated by a short example:

/* First include all needed cygwin stuff. */
#ifdef __CYGWIN__
#include <windows.h>
#include <sys/cygwin.h>
/* Use the following define to determine the Windows version */
#define is_winnt        (GetVersion() < 0x80000000)
#endif

[...]

  struct passwd *user_pwd_entry = getpwnam (username);
  char *cleartext_password = getpass ("Password:");

[...]

#ifdef __CYGWIN__
  /* Patch the typical password test. */
  if (is_winnt)
    {
      HANDLE token;

      /* Try to get the access token from NT. */
      token = cygwin_logon_user (user_pwd_entry, cleartext_password);
      if (token == INVALID_HANDLE_VALUE)
         error_exit;
      /* Inform Cygwin about the new impersonation token.
         Cygwin is able now, to switch to that user context by
         setuid or seteuid calls. */
      cygwin_set_impersonation_token (token);
    }
  else
#endif /* CYGWIN */
    /* Use standard method for W9X as well. */
    hashed_password = crypt (cleartext_password, salt);
    if (!user_pwd_entry ||
        strcmp (hashed_password, user_pwd_entry->pw_password))
      error_exit;

[...]

  /* Everything else remains the same! */

  setegid (user_pwd_entry->pw_gid);
  seteuid (user_pwd_entry->pw_uid);
  execl ("/bin/sh", ...);

The new Cygwin call to retrive an access token is defined as follows:

#include <windows.h>
#include <sys/cygwin.h>

HANDLE
cygwin_logon_user (struct passwd *pw, const char *cleartext_password)

You can call that function as often as you want for different user logons and remeber the access tokens for further calls to the second function.

#include <windows.h>
#include <sys/cygwin.h>

void
cygwin_set_impersonation_token (HANDLE hToken);

is the call to inform Cygwin about the user context to which further calls to setuid/seteuid should switch to. While you need always the correct access token to do a setuid/seteuid to another users context, you are always able to use setuid/seteuid to return to your own user context by giving your own uid as parameter.

If you have remembered several access tokens from calls to cygwin_logon_user you can switch to different user contexts by observing the following order:

  cygwin_set_impersonation_token (user1_token);
  seteuid (user1_uid);

[...]

  seteuid (own_uid);
  cygwin_set_impersonation_token (user2_token);
  seteuid (user2_uid);

[...]

  seteuid (own_uid);
  cygwin_set_impersonation_token (user1_token);
  seteuid (user1_uid);

etc.

2.4.8. New since Cygwin release 1.3.3

Since Cygwin release 1.3.3, applications having the Create a process level token user right can switch user context without giving a password by just calling the usual setuid, seteuid, setgid and setegid functions. This is typically only given to the SYSTEM user. However, this now allows to switch the user context using e. g. rhosts authentication or (when running sshd under SYSTEM account as service) public key authentication.

An important restriction of this method is, that a process started under SYSTEM account can't access network shares which require authentication. This also applies to the subprocesses which switched the user context without a password. People using network home drives are typically not able to access it when trying to login using ssh or rsh without password.

2.4.9. Special values of user and group ids

If the current user is not present in /etc/passwd, that user's user id is set to a special value of 400. The user name for the current user will always be shown correctly. If another user (or a Windows group, treated as a user) is not present in /etc/passwd, the user id of that user will have a special value of -1 (which would be shown by ls as 65535). The user name shown in this case will be '????????'.

If the current user is not present in /etc/passwd, that user's login group id is set to a special value of 401. If another user is not present in /etc/passwd, that user's login group id is set to a special value of -1. If the user is present in /etc/passwd, but that user's group is not in /etc/group and is not the login group of that user, the group id is set to a special value of -1. The name of this group (id -1) will be shown as '????????'. In releases of Cygwin before 1.3.20, the group id 401 had a group name 'None'. Since Cygwin release 1.3.20, the group id 401 is shown as 'mkpasswd', indicating the command that should be run to alleviate the situation.

Also, since Cygwin release 1.3.20, if the current user is present in /etc/passwd, but that user's login group is not present in /etc/group, the group name will be shown as 'mkgroup', again indicating the appropriate command.

To summarize:

• If the current user doesn't show up in /etc/passwd, it's group will be named 'mkpasswd'.

• Otherwise, if the login group of the current user isn't in /etc/group, it will be named 'mkgroup'.

• Otherwise a group not in /etc/group will be shown as '????????' and a user not in /etc/passwd will be shown as "????????".

Note that, since the special user and group names are just indicators, nothing prevents you from actually having a user named `mkpasswd' in /etc/passwd (or a group named `mkgroup' in /etc/group). If you do that, however, be aware of the possible confusion.

3.1.1. Introduction

Cygwin supports both Win32- and POSIX-style paths, where directory delimiters may be either forward or back slashes. UNC pathnames (starting with two slashes and a network name) are also supported.

POSIX operating systems (such as Linux) do not have the concept of drive letters. Instead, all absolute paths begin with a slash (instead of a drive letter such as "c:") and all file systems appear as subdirectories (for example, you might buy a new disk and make it be the /disk2 directory).

Because many programs written to run on UNIX systems assume the existance of a single unified POSIX file system structure, Cygwin maintains a special internal POSIX view of the Win32 file system that allows these programs to successfully run under Windows. Cygwin uses this mapping to translate between Win32 and POSIX paths as necessary.

3.1.2. The Cygwin Mount Table

The mount utility program is used to to map Win32 drives and network shares into Cygwin's internal POSIX directory tree. This is a similar concept to the typical UNIX mount program. For those people coming from a Windows background, the mount utility is very similar to the old DOS join, in that it makes your drive letters appear as subdirectories somewhere else.

The mapping is stored in the current user's Cygwin mount table in the Windows registry so that the information will be retrieved next time the user logs in. Because it is sometimes desirable to have system-wide as well as user-specific mounts, there is also a system-wide mount table that all Cygwin users inherit. The system-wide table may only be modified by a user with the appropriate priviledges (Administrator priviledges in Windows NT).

The current user's table is located under "HKEY_CURRENT_USER/Software/Cygnus Solutions/Cygwin/mounts v<version>" where <version> is the latest registry version associated with the Cygwin library (this version is not the same as the release number). The system-wide table is located under the same subkeys under HKEY_LOCAL_SYSTEM.

By default, the POSIX root / points to the system partition but it can be relocated to any directory in the Windows file system using the mount command. Whenever Cygwin generates a POSIX path from a Win32 one, it uses the longest matching prefix in the mount table. Thus, if C: is mounted as /c and also as /, then Cygwin would translate C:/foo/bar to /c/foo/bar.

Invoking mount without any arguments displays Cygwin's current set of mount points. In the following example, the C drive is the POSIX root and D drive is mapped to /d. Note that in this case, the root mount is a system-wide mount point that is visible to all users running Cygwin programs, whereas the /d mount is only visible to the current user.

c:\> mount
f:\cygwin\bin on /usr/bin type system (binmode)
f:\cygwin\lib on /usr/lib type system (binmode)
f:\cygwin on / type system (binmode)
e:\src on /usr/src type system (binmode)
c: on /cygdrive/c type user (binmode,noumount)
e: on /cygdrive/e type user (binmode,noumount)

You can also use the mount command to add new mount points, and the umount to delete them. See Section 3.6.8 and Section 3.6.15 for more information on how to use these utilities to set up your Cygwin POSIX file system.

Whenever Cygwin cannot use any of the existing mounts to convert from a particular Win32 path to a POSIX one, Cygwin will automatically default to an imaginary mount point under the default POSIX path /cygdrive. For example, if Cygwin accesses Z:\foo and the Z drive is not currently in the mount table, then Z:\ would be automatically converted to /cygdrive/Z. The default prefix of /cygdrive may be changed (see the Section 3.6.8 for more information).

It is possible to assign some special attributes to each mount point. Automatically mounted partitions are displayed as "auto" mounts. Mounts can also be marked as either "textmode" or "binmode" -- whether text files are read in the same manner as binary files by default or not (see Section 3.2 for more information on text and binary modes.

3.1.3. Additional Path-related Information

The cygpath program provides the ability to translate between Win32 and POSIX pathnames in shell scripts. See Section 3.6.2 for the details.

The HOME, PATH, and LD_LIBRARY_PATH environment variables are automatically converted from Win32 format to POSIX format (e.g. from c:\cygwin\bin to /bin, if there was a mount from that Win32 path to that POSIX path) when a Cygwin process first starts.

Symbolic links can also be used to map Win32 pathnames to POSIX. For example, the command ln -s //pollux/home/joe/data /data would have about the same effect as creating a mount point from //pollux/home/joe/data to /data using mount, except that symbolic links cannot set the default file access mode. Other differences are that the mapping is distributed throughout the file system and proceeds by iteratively walking the directory tree instead of matching the longest prefix in a kernel table. Note that symbolic links will only work on network drives that are properly configured to support the "system" file attribute. Many do not do so by default (the Unix Samba server does not by default, for example).

3.2.1. The Issue

On a UNIX system, when an application reads from a file it gets exactly what's in the file on disk and the converse is true for writing. The situation is different in the DOS/Windows world where a file can be opened in one of two modes, binary or text. In the binary mode the system behaves exactly as in UNIX. However in text mode there are major differences:

1. On writing in text mode, a NL (\n, ^J) is transformed into the sequence CR (\r, ^M) NL.

2. On reading in text mode, a CR followed by an NL is deleted and a ^Z character signals the end of file.

This can wreak havoc with the seek/fseek calls since the number of bytes actually in the file may differ from that seen by the application.

The mode can be specified explicitly as explained in the Programming section below. In an ideal DOS/Windows world, all programs using lines as records (such as bash, make, sed ...) would open files (and change the mode of their standard input and output) as text. All other programs (such as cat, cmp, tr ...) would use binary mode. In practice with Cygwin, programs that deal explicitly with object files specify binary mode (this is the case of od, which is helpful to diagnose CR problems). Most other programs (such as cat, cmp, tr) use the default mode.

3.2.2. The default Cygwin behavior

The Cygwin system gives us some flexibility in deciding how files are to be opened when the mode is not specified explicitly. The rules are evolving, this section gives the design goals.

1. If the file appears to reside on a file system that is mounted (i.e. if its pathname starts with a directory displayed by mount), then the default is specified by the mount flag. If the file is a symbolic link, the mode of the target file system applies.

2. If the file appears to reside on a file system that is not mounted (as can happen when the path contains a drive letter), the default is text.

3. Pipes and non-file devices are opened in binary mode, except if the CYGWIN environment variable contains nobinmode.

   Warning!

   In b20.1 of 12/98, a file will be opened in binary mode if any of the following conditions hold: binary mode is specified in the open call CYGWIN contains binmode the file resides in a binary mounted partition the file is not a disk file

4. When a Cygwin program is launched by a shell, its standard input, output and error are in binary mode if the CYGWIN variable contains tty, else in text mode, except if they are piped or redirected.

   When redirecting, the Cygwin shells uses rules (a-c). For these shells the relevant value of CYGWIN is that at the time the shell was launched and not that at the time the program is executed. Non-Cygwin shells always pipe and redirect with binary mode. With non-Cygwin shells the commands cat filename | program and program < filename are not equivalent when filename is on a text-mounted partition.

3.2.3. Example

To illustrate the various rules, we provide scripts to delete CRs from files by using the tr program, which can only write to standard output. The script

#!/bin/sh
# Remove \r from the file given as argument
tr -d '\r' < "$1" > "$1".nocr

will not work on a text mounted systems because the \r will be reintroduced on writing. However scripts such as

#!/bin/sh
# Remove \r from the file given as argument
tr -d '\r' | gzip | gunzip > "$1".nocr

and the .bat file

REM Remove \r from the file given as argument
@echo off
tr -d \r < %1 > %1.nocr

work fine. In the first case (assuming the pipes are binary) we rely on gunzip to set its output to binary mode, possibly overriding the mode used by the shell. In the second case we rely on the DOS shell to redirect in binary mode.

3.2.4. Binary or text?

UNIX programs that have been written for maximum portability will know the difference between text and binary files and act appropriately under Cygwin. For those programs, the text mode default is a good choice. Programs included in official Cygwin distributions should work well in the default mode.

Text mode makes it much easier to mix files between Cygwin and Windows programs, since Windows programs will usually use the CRLF format. Unfortunately you may still have some problems with text mode. First, some of the utilities included with Cygwin do not yet specify binary mode when they should, e.g. cat will not work with binary files (input will stop at ^Z, CRs will be introduced in the output). Second, you will introduce CRs in text files you write, which can cause problems when moving them back to a UNIX system.

If you are mounting a remote file system from a UNIX machine, or moving files back and forth to a UNIX machine, you may want to access the files in binary mode. The text files found there will normally be in UNIX NL format, and you would want any files put there by Cygwin programs to be stored in a format understood by UNIX. Be sure to remove CRs from all Makefiles and shell scripts and make sure that you only edit the files with DOS/Windows editors that can cope with and preserve NL terminated lines.

Note that you can decide this on a disk by disk basis (for example, mounting local disks in text mode and network disks in binary mode). You can also partition a disk, for example by mounting c: in text mode, and c:\home in binary mode.

3.2.5. Programming

In the open() function call, binary mode can be specified with the flag O_BINARY and text mode with O_TEXT. These symbols are defined in fcntl.h.

In the fopen() function call, binary mode can be specified by adding a b to the mode string. There is no direct way to specify text mode.

The mode of a file can be changed by the call setmode(fd,mode) where fd is a file descriptor (an integer) and mode is O_BINARY or O_TEXT. The function returns O_BINARY or O_TEXT depending on the mode before the call, and EOF on error.

3.4.1. DOS devices

Windows filenames invalid under Windows are also invalid under Cygwin. This means that base filenames such as AUX, COM1, LPT1 or PRN (to name a few) cannot be used in a regular Cygwin Windows or POSIX path, even with an extension (prn.txt). However the special names can be used as filename extensions (file.aux). You can use the special names as you would under DOS, for example you can print on your default printer with the command cat filename > PRN (make sure to end with a Form Feed).

3.4.2. POSIX devices

There is no need to create a POSIX /dev directory as it is simulated within Cygwin automatically. It supports the following devices: /dev/null, /dev/zero, /dev/tty, /dev/ttyX, /dev/ptmx, /dev/comX (the serial ports), /dev/windows (the windows message queue), /dev/random and /dev/urandom. These devices cannot be seen with the command ls /dev although commands such as ls /dev/tty work fine.

Windows NT/W2K/XP additionally support raw devices like floppies, disks, partitions and tapes. These are accessed from Cygwin applications using POSIX device names which are supported in two different ways. Up to 1.3.3, Cygwin only uses Win32 device names, since 1.3.4 it additionally uses NT internal device names.

Up to Cygwin 1.3.3 the only way to access those devices is to mount the Win32 device names to a POSIX device name.

The Win32 device name for a partition is the drive letter with leading \\.\, so the floppy would be \\.\A:, the first partition typically \\.\C:. Complete drives (except floppies and CD-ROMS which are supported as partitions only) are named \\.\PHYSICALDRIVEx. The x is the drive number which you can check in the disk manager. Each drive line has prepended the text "Disk x".

To access tape drives the Win32 file name \\.\TAPEx is used. For example the first installed tape device is named \\.\tape0.

The naming convention is simple: The name of the POSIX device has to begin with /dev/ and the rest is as you like. The only exception are tape devices. To identify if the tape device is used as a rewind or a no-rewind device the name must not begin with n (rewind) or has to begin with n (no-rewind).

Some examples:

mount -b //./A: /dev/fd0              # mount floppy as raw block special
mount -b //./physicaldrive1 /dev/hdb  # mount "Disk 1"
mount -b //./tape0 /dev/st0           # mount first tape as the rewind device...
mount -b //./tape0 /dev/nst0          # ...and as the no-rewind device

Note the usage of the -b option. It is best to include the -b option when mounting these devices to ensure that all file I/O is in "binary mode".

Since Cygwin 1.3.4 raw devices are accessible from inside of Cygwin processes using fixed POSIX device names. That means, you don't have to mount the devices anymore which results in a more cleaner mount table.

These new fixed POSIX device names are generated using a direct conversion from the POSIX namespace to the internal NT namespace. E.g. the first harddisk is the NT internal device \device\harddisk0\partition0 or the first partition on the third harddisk is \device\harddisk2\partition1. The first floppy in the system is \device\floppy0, the first CD-ROM is \device\cdrom0 and the first tape drive is \device\tape0.

The new fixed POSIX names are mapped to NT internal devices as follows:

/dev/st0	\device\tape0, rewind
/dev/nst0	\device\tape0, no-rewind
/dev/st1	\device\tape1
...

/dev/fd0	\device\floppy0
/dev/fd1	\device\floppy1
...

/dev/scd0	\device\cdrom0
/dev/scd1	\device\cdrom1
...

/dev/sda	\device\harddisk0\partition0	(whole disk)
/dev/sda1	\device\harddisk0\partition1	(first partition)
...
/dev/sda15	\device\harddisk0\partition15	(fifteenth partition)

/dev/sdb	\device\harddisk1\partition0
/dev/sdb1	\device\harddisk1\partition1

[up to]

/dev/sdl	\device\harddisk11\partition0
/dev/sdl1	\device\harddisk11\partition1
...
/dev/sdl15	\device\harddisk11\partition15

if you don't like these device names, feel free to create symbolic links as they are created on Linux systems for convenience:

ln -s /dev/scd0 /dev/cdrom
ln -s /dev/st0  /dev/tape
...

Note that you can't use the mount table to map from fixed device name to your own device name or to map from internal NT device name to your own device name. The following two examples will not work:

mount -s -f -b /dev/st0 /dev/tape
mount -s -f -b /device/tape0 /dev/tape

3.4.3. The .exe extension

Executable program filenames end with .exe but the .exe need not be included in the command, so that traditional UNIX names can be used. However, for programs that end in ".bat" and ".com", you cannot omit the extension.

As a side effect, the ls filename gives information about filename.exe if filename.exe exists and filename does not. In the same situation the function call stat("filename",..) gives information about filename.exe. The two files can be distinguished by examining their inodes, as demonstrated below.

C:\> ls * 
a      a.exe     b.exe
C:\> ls -i a a.exe
445885548 a       435996602 a.exe
C:\> ls -i b b.exe
432961010 b       432961010 b.exe

The gcc compiler produces an executable named filename.exe when asked to produce filename. This allows many makefiles written for UNIX systems to work well under Cygwin.

Unfortunately, the install and strip commands do distinguish between filename and filename.exe. They fail when working on a non-existing filename even if filename.exe exists, thus breaking some makefiles. This problem can be solved by writing install and strip shell scripts to provide the extension ".exe" when needed.

3.4.4. The @pathnames

To circumvent the limitations on shell line length in the native Windows command shells, Cygwin programs expand their arguments starting with "@" in a special way. If a file pathname exists, the argument @pathname expands recursively to the content of pathname. Double quotes can be used inside the file to delimit strings containing blank space. Embedded double quotes must be repeated. In the following example compare the behaviors of the bash built-in echo and of the program /bin/echo.

bash$ echo  'This   is   "a     long"  line' > mylist
bash$ echo @mylist
@mylist
c:\> c:\cygwin\bin\echo @mylist
This is a     long line

3.6.1. cygcheck

Usage: cygcheck [OPTIONS] [program ...]
Check system information or PROGRAM library dependencies

 -c, --check-setup  check packages installed via setup.exe
 -s, --sysinfo      system information (not with -k)
 -v, --verbose      verbose output (indented) (for -s or programs)
 -r, --registry     registry search (requires -s)
 -k, --keycheck     perform a keyboard check session (not with -s)
 -h, --help         give help about the info (not with -c)
 -V, --version      output version information and exit
You must at least give either -s or -k or a program name

The cygcheck program is a diagnostic utility that examines your system and reports the information that is significant to the proper operation of Cygwin programs. It can give information about specific programs or libraries you are trying to run, general system information, or both. If you list one or more programs on the command line, it will diagnose the runtime environment of that program or programs, providing the names of DLL files on which the program depends. If you specify the -s option, it will give general system information. If you specify -s and list one or more programs on the command line, it reports on both.

The -h option prints additional helpful messages in the report, at the beginning of each section. It also adds table column headings. While this is useful information, it also adds some to the size of the report, so if you want a compact report or if you know what everything is already, just leave this out.

The -v option causes the output to be more verbose. What this means is that additional information will be reported which is usually not interesting, such as the internal version numbers of DLLs, additional information about recursive DLL usage, and if a file in one directory in the PATH also occurs in other directories on the PATH.

The -r option causes cygcheck to search your registry for information that is relevent to Cygwin programs. These registry entries are the ones that have "Cygwin" in the name. If you are paranoid about privacy, you may remove information from this report, but please keep in mind that doing so makes it harder to diagnose your problems.

The -c option causes the arguments to be interpreted as package names. cygcheck will report the current version of the package that you specify, or with no arguments, on all packages.

The cygcheck program should be used to send information about your system for troubleshooting when requested. When asked to run this command save the output so that you can email it, for example:

C:\cygwin> cygcheck -s -v -r -h > cygcheck_output.txt

3.6.2. cygpath

Usage: cygpath (-d|-m|-u|-w|-t TYPE) [-c HANDLE] [-f FILE] [options] NAME
       cygpath [-ADHPSW] 
Convert Unix and Windows format paths, or output system path information

Output type options:
  -d, --dos             print DOS (short) form of NAME (C:\PROGRA~1\)
  -m, --mixed           like --windows, but with regular slashes (C:/WINNT)
  -u, --unix            (default) print Unix form of NAME (/cygdrive/c/winnt)
  -w, --windows         print Windows form of NAME (C:\WINNT)
  -t, --type TYPE       print TYPE form: 'dos', 'mixed', 'unix', or 'windows'
Path conversion options:
  -a, --absolute        output absolute path
  -l, --long-name       print Windows long form of NAME (with -w, -m only)
  -p, --path            NAME is a PATH list (i.e., '/bin:/usr/bin')
  -s, --short-name      print DOS (short) form of NAME (with -w, -m only)
System information:
  -A, --allusers        use `All Users' instead of current user for -D, -P
  -D, --desktop         output `Desktop' directory and exit
  -H, --homeroot        output `Profiles' directory (home root) and exit
  -P, --smprograms      output Start Menu `Programs' directory and exit
  -S, --sysdir          output system directory and exit
  -W, --windir          output `Windows' directory and exit

The cygpath program is a utility that converts Windows native filenames to Cygwin POSIX-style pathnames and vice versa. It can be used when a Cygwin program needs to pass a file name to a native Windows program, or expects to get a file name from a native Windows program. Alternatively, cygpath can output information about the location of important system directories in either format.

The -u and -w options indicate whether you want a conversion to UNIX (POSIX) format (-u) or to Windows format (-w). Use the -d to get DOS-style (8.3) file and path names. The -m option will output Windows-style format but with forward slashes instead of backslashes. This option is especially useful in shell scripts, which use backslashes as an escape character.

In combination with the -w option, you can use the -l and -s options to use normal (long) or DOS-style (short) form. The -d option is identical to -w and -s together.

Caveat: The -l option does not work if the check_case parameter of CYGWIN is set to strict, since Cygwin is not able to match any Windows short path in this mode.

The -p option means that you want to convert a path-style string rather than a single filename. For example, the PATH environment variable is semicolon-delimited in Windows, but colon-delimited in UNIX. By giving -p you are instructing cygpath to convert between these formats.

The -i option supresses the print out of the usage message if no filename argument was given. It can be used in make file rules converting variables that may be omitted to a proper format. Note that cygpath output may contain spaces (C:\Program Files) so should be enclosed in quotes.

#!/bin/sh
if [ "${1}" = "" ];
	then
		XPATH=".";
	else
		XPATH="$(cygpath -w "${1}")";
	fi
	explorer $XPATH &

The capital options -D, -H, -P, -S, and -W output directories used by Windows that are not the same on all systems, for example -S might output C:\WINNT\SYSTEM32 or C:\WINDOWS\SYSTEM. The -H shows the Windows profiles directory that can be used as root of home. The -A option forces use of the "All Users" directories instead of the current user for the -D and -P options. On Win9x systems with only a single user, -A has no effect; -D and -AD would have the same output. By default the output is in UNIX (POSIX) format; use the -w or -d options to get other formats.

3.6.3. dumper

Usage: dumper [OPTION] FILENAME WIN32PID
Dump core from WIN32PID to FILENAME.core

-d, --verbose  be verbose while dumping
-h, --help     output help information and exit
-q, --quiet    be quiet while dumping (default)
-v, --version  output version information and exit

The dumper utility can be used to create a core dump of running Windows process. This core dump can be later loaded to gdb and analyzed. One common way to use dumper is to plug it into cygwin's Just-In-Time debugging facility by adding

error_start=x:\path\to\dumper.exe

dumper can be also be started from the command line to create a core dump of any running process. Unfortunately, because of a Windows API limitation, when a core dump is created and dumper exits, the target process is terminated too.

To save space in the core dump, dumper doesn't write those portions of target process' memory space that are loaded from executable and dll files and are unchangeable, such as program code and debug info. Instead, dumper saves paths to files which contain that data. When a core dump is loaded into gdb, it uses these paths to load appropriate files. That means that if you create a core dump on one machine and try to debug it on another, you'll need to place identical copies of the executable and dlls in the same directories as on the machine where the core dump was created.

3.6.4. getfacl

Usage: getfacl [-adn] FILE [FILE2...]
Display file and directory access control lists (ACLs).

  -a, --all      display the filename, the owner, the group, and
                 the ACL of the file
  -d, --dir      display the filename, the owner, the group, and
                 the default ACL of the directory, if it exists
  -h, --help     output usage information and exit
  -n, --noname   display user and group IDs instead of names
  -v, --version  output version information and exit

When multiple files are specified on the command line, a blank
line separates the ACLs for each file.

For each argument that is a regular file, special file or directory, getfacl displays the owner, the group, and the ACL. For directories getfacl displays additionally the default ACL. With no options specified, getfacl displays the filename, the owner, the group, and both the ACL and the default ACL, if it exists. For more information on Cygwin and Windows ACLs, see see Section 2.4 in the Cygwin User's Guide. The format for ACL output is as follows:

     # file: filename
     # owner: name or uid
     # group: name or uid
     user::perm
     user:name or uid:perm
     group::perm
     group:name or gid:perm
     mask:perm
     other:perm
     default:user::perm
     default:user:name or uid:perm
     default:group::perm
     default:group:name or gid:perm
     default:mask:perm
     default:other:perm

3.6.5. kill

Usage: kill [-f] [-signal] [-s signal] pid1 [pid2 ...]
       kill -l [signal]
Send signals to processes

 -f, --force     force, using win32 interface if necessary
 -l, --list      print a list of signal names
 -s, --signal    send signal (use kill --list for a list)
 -h, --help      output usage information and exit
 -v, --version   output version information and exit

The kill program allows you to send arbitrary signals to other Cygwin programs. The usual purpose is to end a running program from some other window when ^C won't work, but you can also send program-specified signals such as SIGUSR1 to trigger actions within the program, like enabling debugging or re-opening log files. Each program defines the signals they understand.

You may need to specify the full path to use kill from within some shells, including bash, the default Cygwin shell. This is because bash defines a kill builtin function; see the bash man page under BUILTIN COMMANDS for more information. To make sure you are using the Cygwin version, try

$ /bin/kill --version

Unless you specific the -f option, the "pid" values used by kill are the Cygwin pids, not the Windows pids. To get a list of running programs and their Cygwin pids, use the Cygwin ps program. ps -W will display all windows pids.

The kill -l option prints the name of the given signal, or a list of all signal names if no signal is given.

To send a specific signal, use the -signN option, either with a signal number or a signal name (minus the "SIG" part), like these examples:

$ kill 123
$ kill -1 123
$ kill -HUP 123
$ kill -f 123

Here is a list of available signals, their numbers, and some commentary on them, from the file <sys/signal.h>, which should be considered the official source of this information.

SIGHUP       1    hangup
SIGINT       2    interrupt
SIGQUIT      3    quit
SIGILL       4    illegal instruction (not reset when caught)
SIGTRAP      5    trace trap (not reset when caught)
SIGABRT      6    used by abort
SIGEMT       7    EMT instruction
SIGFPE       8    floating point exception
SIGKILL      9    kill (cannot be caught or ignored)
SIGBUS      10    bus error
SIGSEGV     11    segmentation violation
SIGSYS      12    bad argument to system call
SIGPIPE     13    write on a pipe with no one to read it
SIGALRM     14    alarm clock
SIGTERM     15    software termination signal from kill
SIGURG      16    urgent condition on IO channel
SIGSTOP     17    sendable stop signal not from tty
SIGTSTP     18    stop signal from tty
SIGCONT     19    continue a stopped process
SIGCHLD     20    to parent on child stop or exit
SIGTTIN     21    to readers pgrp upon background tty read
SIGTTOU     22    like TTIN for output if (tp->t_local&LTOSTOP)
SIGPOLL     23    System V name for SIGIO
SIGXCPU     24    exceeded CPU time limit
SIGXFSZ     25    exceeded file size limit
SIGVTALRM   26    virtual time alarm
SIGPROF     27    profiling time alarm
SIGWINCH    28    window changed
SIGLOST     29    resource lost (eg, record-lock lost)
SIGUSR1     30    user defined signal 1
SIGUSR2     31    user defined signal 2

3.6.6. mkgroup

Usage: mkgroup [OPTION]... [domain]...
Prints /etc/group file to stdout

Options:
   -l,--local             print local group information
   -c,--current           print current group, if a domain account
   -d,--domain            print global group information (from current
                          domain if no domains specified).
   -o,--id-offset offset  change the default offset (10000) added to gids
                          in domain accounts.
   -s,--no-sids           don't print SIDs in pwd field
                          (this affects ntsec)
   -u,--users             print user list in gr_mem field
   -h,--help              print this message

   -v,--version           print version information and exit

One of `-l' or `-d' must be given on NT/W2K.

The mkgroup program can be used to help configure your Windows system to be more UNIX-like by creating an initial /etc/group. Its use is essential on the NT series (Windows NT, 2000, and XP) to include Windows security information. It can also be used on the Win9x series (Windows 95, 98, and Me) to create a file with the correct format. To initially set up your machine if you are a local user, you'd do something like this:

$ mkdir /etc
$ mkgroup -l > /etc/group

Note that this information is static. If you change the group information in your system, you'll need to regenerate the group file for it to have the new information.

The -d and -l options allow you to specify where the information comes from, the local machine or the domain (default or given), or both. With the -d option the program contacts the Domain Controller, which my be unreachable or have restricted access. An entry for the current domain user can then be created by using the option -c together with -l, but -c has no effect when used with -d. The -o option allows for special cases (such as multiple domains) where the GIDs might match otherwise. The -s option omits the NT Security Identifier (SID). For more information on SIDs, see Section 2.4 in the Cygwin User's Guide. The -u option causes mkgroup to enumerate the users for each group, placing the group members in the gr_mem (last) field. Note that this can greatly increase the time for mkgroup to run in a large domain. Having gr_mem fields is helpful when a domain user logs in remotely while the local machine is disconnected from the Domain Controller.

3.6.7. mkpasswd

Usage: mkpasswd [OPTION]... [domain]...
Prints /etc/passwd file to stdout

Options:
   -l,--local              print local user accounts
   -c,--current            print current account, if a domain account
   -d,--domain             print domain accounts (from current domain
                           if no domains specified)
   -o,--id-offset offset   change the default offset (10000) added to uids
                           in domain accounts.
   -g,--local-groups       print local group information too
                           if no domains specified
   -m,--no-mount           don't use mount points for home dir
   -s,--no-sids            don't print SIDs in GCOS field
                           (this affects ntsec)
   -p,--path-to-home path  use specified path and not user account home dir or /home
   -u,--username username  only return information for the specified user
   -h,--help               displays this message
   -v,--version            version information and exit

One of `-l', `-d' or `-g' must be given on NT/W2K.

The mkpasswd program can be used to help configure your Windows system to be more UNIX-like by creating an initial /etc/passwd from your system information. Its use is essential on the NT series (Windows NT, 2000, and XP) to include Windows security information, but the actual passwords are determined by Windows, not by the content of /etc/passwd. On the Win9x series (Windows 95, 98, and Me) the password field must be replaced by the output of crypt your_password if remote access is desired. To initially set up your machine if you are a local user, you'd do something like this:

$ mkdir /etc
$ mkpasswd -l > /etc/passwd

Note that this information is static. If you change the user information in your system, you'll need to regenerate the passwd file for it to have the new information.

The -d and -l options allow you to specify where the information comes from, the local machine or the domain (default or given), or both. With the -d option the program contacts the Domain Controller, which my be unreachable or have restricted access. An entry for the current domain user can then be created by using the option -c together with -l, but -c has no effect when used with -d. The -o option allows for special cases (such as multiple domains) where the UIDs might match otherwise. The -g option creates a local user that corresponds to each local group. This is because NT assigns groups file ownership. The -m option bypasses the current mount table so that, for example, two users who have a Windows home directory of H: could mount them differently. The -s option omits the NT Security Identifier (SID). For more information on SIDs, see Section 2.4 in the Cygwin User's Guide. The -p option causes mkpasswd to use the specified prefix instead of the account home dir or /home/. For example, this command:

$ mkpasswd -l -p "$(cygpath -H)" > /etc/passwd

3.6.8. mount

Usage: mount [OPTION] [<win32path> <posixpath>]
Display information about mounted filesystems, or mount a filesystem

  -b, --binary     (default)    text files are equivalent to binary files
                                (newline = \n)
  -c, --change-cygdrive-prefix  change the cygdrive path prefix to <posixpath>
  -f, --force                   force mount, don't warn about missing mount
                                point directories
  -h, --help                    output usage information and exit
  -m, --mount-commands          write mount commands to replace user and
                                system mount points and cygdrive prefixes
  -o, --options X[,X...]        specify mount options
  -p, --show-cygdrive-prefix    show user and/or system cygdrive path prefix
  -s, --system     (default)    add system-wide mount point
  -t, --text                    text files get \r\n line endings
  -u, --user                    add user-only mount point
  -v, --version                 output version information and exit
  -x, --executable              treat all files under mount point as executables
  -E, --no-executable           treat all files under mount point as 
                                non-executables
  -X, --cygwin-executable       treat all files under mount point as cygwin
                                executables

The mount program is used to map your drives and shares onto Cygwin's simulated POSIX directory tree, much like as is done by mount commands on typical UNIX systems. Please see Section 3.1.2 for more information on the concepts behind the Cygwin POSIX file system and strategies for using mounts. To remove mounts, use umount

c:\cygwin\> mount
c:\cygwin\bin on /usr/bin type system (binmode)
c:\cygwin\lib on /usr/lib type system (binmode)
c:\cygwin on / type system (binmode)
c: on /c type user (binmode,noumount)
d: on /d type user (binmode,noumount)

c:\cygwin\> ls /data
ls: /data: No such file or directory
c:\cygwin\> mount \\pollux\home\joe\data /data
mount: warning - /data does not exist!
c:\cygwin\> mount
\\pollux\home\joe\data on /data type sytem (binmode)
c:\cygwin\bin on /usr/bin type system (binmode)
c:\cygwin\lib on /usr/lib type system (binmode)
c:\cygwin on / type system (binmode)
c: on /c type user (binmode,noumount)
d: on /d type user (binmode,noumount)

  user       - mount lives user-specific mount
  system     - mount lives in system table (default)
  binary     - files default to binary mode (default)
  text       - files default to CRLF text mode line endings
  exec       - files below mount point are all executable
  notexec    - files below mount point are not executable
  cygexec    - files below mount point are all cygwin executables
  nosuid     - no suid files are allowed (currently unimplemented)
  managed    - directory is managed by cygwin.  Mixed case and special
               characters in filenames are allowed.

c:\cygwin\> mount --change-cygdrive-prefix /

3.6.9. passwd

Usage: passwd (-l|-u|-S) [USER]
       passwd [-i NUM] [-n MINDAYS] [-x MAXDAYS] [-L LEN]
Change USER's password or password attributes

User operations:
 -l, --lock      lock USER's account
 -u, --unlock    unlock USER's account
 -S, --status    display password status for USER (locked, expired, etc.)

System operations:
 -i, --inactive  set NUM of days before inactive accounts are disabled
                 (inactive accounts are those with expired passwords)
 -n, --minage    set system minimum password age to MINDAYS
 -x, --maxage    set system maximum password age to MAXDAYS
 -L, --length    set system minimum password length to LEN

Other options:
 -h, --help      output usage information and exit
 -v, --version   output version information and exit

passwd changes passwords for user accounts. A normal user may only change the password for their own account, but administrators may change passwords on any account. passwd also changes account information, such as password expiry dates and intervals.

Password changes: The user is first prompted for their old password, if one is present. This password is then encrypted and compared against the stored password. The user has only one chance to enter the correct password. The administrators are permitted to bypass this step so that forgotten passwords may be changed.

The user is then prompted for a replacement password. passwd will prompt twice for this replacement and compare the second entry against the first. Both entries are require to match in order for the password to be changed.

After the password has been entered, password aging information is checked to see if the user is permitted to change their password at this time. If not, passwd refuses to change the password and exits.

Account maintenance: User accounts may be locked and unlocked with the -l and -u flags. The -l option disables an account. The -u option re-enables an account.

The account status may be given with the -S option. The status information is self explanatory.

Administrators can also use passwd to change system-wide password expiry and length requirements with the -i, -n, -x, and -L options. The -i option is used to disable an account after the password has been expired for a number of days. After a user account has had an expired password for NUM days, the user may no longer sign on to the account. The -n option is used to set the minimum number of days before a password may be changed. The user will not be permitted to change the password until MINDAYS days have elapsed. The -x option is used to set the maximum number of days a password remains valid. After MAXDAYS days, the password is required to be changed. Allowed values for the above options are 0 to 999. The -L option sets the minimum length of allowed passwords for users who don't belong to the administrators group to LEN characters. Allowed values for the minimum password length are 0 to 14. In any of the above cases, a value of 0 means `no restrictions'.

Limitations: Users may not be able to change their password on some systems.

3.6.10. ps

Usage: ps [-aefls] [-u UID]
Report process status

 -a, --all       show processes of all users
 -e, --everyone  show processes of all users
 -f, --full      show process uids, ppids
 -h, --help      output usage information and exit
 -l, --long      show process uids, ppids, pgids, winpids
 -s, --summary   show process summary
 -u, --user      list processes owned by UID
 -v, --version   output version information and exit
 -W, --windows   show windows as well as cygwin processes
With no options, ps outputs the long format by default

The ps program gives the status of all the Cygwin processes running on the system (ps = "process status"). Due to the limitations of simulating a POSIX environment under Windows, there is little information to give.

The PID column is the process ID you need to give to the kill command. The PPID is the parent process ID, and PGID is the process group ID. The WINPID column is the process ID displayed by NT's Task Manager program. The TTY column gives which pseudo-terminal a process is running on, or a '?' for services. The UID column shows which user owns each process. STIME is the time the process was started, and COMMAND gives the name of the program running.

By default ps will only show processes owned by the current user. With either the -a or -e option, all user's processes (and system processes) are listed. There are historical UNIX reasons for the synonomous options, which are functionally identical. The -f option outputs a "full" listing with usernames for UIDs. The -l option is the default display mode, showing a "long" listing with all the above columns. The other display option is -s, which outputs a shorter listing of just PID, TTY, STIME, and COMMAND. The -u option allows you to show only processes owned by a specific user. The -W option causes ps show non-Cygwin Windows processes as well as Cygwin processes. The WINPID is also the PID, and they can be killed with the Cygwin kill command's -f option.

3.6.11. regtool

Usage: regtool.exe [OPTION] (add | check | get | list | remove | unset) KEY
View or edit the Win32 registry

Actions:
 add KEY\SUBKEY             add new SUBKEY
 check KEY                  exit 0 if KEY exists, 1 if not
 get KEY\VALUE              prints VALUE to stdout
 list KEY                   list SUBKEYs and VALUEs
 remove KEY                 remove KEY
 set KEY\VALUE [data ...]   set VALUE
 unset KEY\VALUE            removes VALUE from KEY

Options for 'list' Action:
 -k, --keys           print only KEYs
 -l, --list           print only VALUEs
 -p, --postfix        like ls -p, appends '\' postfix to KEY names

Options for 'set' Action:
 -e, --expand-string  set type to REG_EXPAND_SZ
 -i, --integer        set type to REG_DWORD
 -m, --multi-string   set type to REG_MULTI_SZ
 -s, --string         set type to REG_SZ

Options for 'set' and 'unset' Actions:
 -K<c>, --key-separator[=]<c>  set key separator to <c> instead of '\'

Other Options:
 -h, --help     output usage information and exit
 -q, --quiet    no error output, just nonzero return if KEY/VALUE missing
 -v, --verbose  verbose output, including VALUE contents when applicable
 -V, --version  output version information and exit

KEY is in the format [host]\prefix\KEY\KEY\VALUE, where host is optional
remote host in either \\hostname or hostname: format and prefix is any of:
  root     HKCR  HKEY_CLASSES_ROOT (local only)
  config   HKCC  HKEY_CURRENT_CONFIG (local only)
  user     HKCU  HKEY_CURRENT_USER (local only)
  machine  HKLM  HKEY_LOCAL_MACHINE
  users    HKU   HKEY_USERS

You can use forward slash ('/') as a separator instead of backslash, in
that case backslash is treated as escape character
Example: regtool.exe get '\user\software\Microsoft\Clock\iFormat'

The regtool program allows shell scripts to access and modify the Windows registry. Note that modifying the Windows registry is dangerous, and carelessness here can result in an unusable system. Be careful.

The -v option means "verbose". For most commands, this causes additional or lengthier messages to be printed. Conversely, the -q option supresses error messages, so you can use the exit status of the program to detect if a key exists or not (for example).

You must provide regtool with an action following options (if any). Currently, the action must be add, set, check, get, list, remove, set, or unset.

The add action adds a new key. The check action checks to see if a key exists (the exit code of the program is zero if it does, nonzero if it does not). The get action gets the value of a value of a key, and prints it (and nothing else) to stdout. Note: if the value doesn't exist, an error message is printed and the program returns a non-zero exit code. If you give -q, it doesn't print the message but does return the non-zero exit code.

The list action lists the subkeys and values belonging to the given key. With list, the -k option instructs regtool to print only KEYs, and the -l option to print only VALUEs. The -p option postfixes a '/' to each KEY, but leave VALUEs with no postfix. The remove action removes a key. Note that you may need to remove everything in the key before you may remove it, but don't rely on this stopping you from accidentally removing too much.

The set action sets a value within a key. -e means it's an expanding string (REG_EXPAND_SZ) that contains embedded environment variables. -i means the value is an integer (REG_DWORD). -m means it's a multi-string (REG_MULTI_SZ). -s means the value is a string (REG_SZ). If you don't specify one of these, regtool tries to guess the type based on the value you give. If it looks like a number, it's a DWORD. If it starts with a percent, it's an expanding string. If you give multiple values, it's a multi-string. Else, it's a regular string. The unset action removes a value from a key.

By default, the last "\" or "/" is assumed to be the separator between the key and the value. You can use the -K option to provide an alternate key/value separator character.

3.6.12. setfacl

Usage: setfacl [-r] (-f ACL_FILE | -s acl_entries) FILE...
       setfacl [-r] ([-d acl_entries] [-m acl_entries]) FILE...
Modify file and directory access control lists (ACLs)

  -d, --delete     delete one or more specified ACL entries
  -f, --file       set ACL entries for FILE to ACL entries read
                   from a ACL_FILE
  -m, --modify     modify one or more specified ACL entries
  -r, --replace    replace mask entry with maximum permissions
                   needed for the file group class
  -s, --substitute substitute specified ACL entries for the
                   ACL of FILE
  -h, --help       output usage information and exit
  -v, --version    output version information and exit

At least one of (-d, -f, -m, -s) must be specified

For each file given as parameter, setfacl will either replace its complete ACL (-s, -f), or it will add, modify, or delete ACL entries. For more information on Cygwin and Windows ACLs, see see Section 2.4 in the Cygwin User's Guide.

Acl_entries are one or more comma-separated ACL entries from the following list:

         u[ser]::perm
         u[ser]:uid:perm
         g[roup]::perm
         g[roup]:gid:perm
         m[ask]::perm
         o[ther]::perm

         d[efault]:u[ser]:uid:perm

perm is either a 3-char permissions string in the form "rwx" with the character '-' for no permission or it is the octal representation of the permissions, a value from 0 (equivalent to "---") to 7 ("rwx"). uid is a user name or a numerical uid. gid is a group name or a numerical gid.

The following options are supported:

-d Delete one or more specified entries from the file's ACL. The owner, group and others entries must not be deleted. Acl_entries to be deleted should be specified without permissions, as in the following list:

         u[ser]:uid
         g[roup]:gid
         d[efault]:u[ser]:uid
         d[efault]:g[roup]:gid
         d[efault]:m[ask]:
         d[efault]:o[ther]:

-f Take the Acl_entries from ACL_FILE one per line. Whitespace characters are ignored, and the character "#" may be used to start a comment. The special filename "-" indicates reading from stdin. Note that you can use this with getfacl and setfacl to copy ACLs from one file to another:

$ getfacl source_file | setfacl -f - target_file

Required entries are: one user entry for the owner of the file, one group entry for the group of the file, and one other entry.

If additional user and group entries are given: a mask entry for the file group class of the file, and no duplicate user or group entries with the same uid/gid.

If it is a directory: one default user entry for the owner of the file, one default group entry for the group of the file, one default mask entry for the file group class, and one default other entry.

-m Add or modify one or more specified ACL entries. Acl_entries is a comma-separated list of entries from the same list as above.

-r Causes the permissions specified in the mask entry to be ignored and replaced by the maximum permissions needed for the file group class.

-s Like -f, but substitute the file's ACL with Acl_entries specified in a comma-separated list on the command line.

While the -d and -m options may be used in the same command, the -f and -s options may be used only exclusively.

Directories may contain default ACL entries. Files created in a directory that contains default ACL entries will have permissions according to the combination of the current umask, the explicit permissions requested and the default ACL entries

Limitations: Under Cygwin, the default ACL entries are not taken into account currently.

3.6.13. ssp

Usage: ssp [options] low_pc high_pc command...
Single-step profile COMMAND

 -c, --console-trace  trace every EIP value to the console. *Lots* slower.
 -d, --disable        disable single-stepping by default; use
                      OutputDebugString ("ssp on") to enable stepping
 -e, --enable         enable single-stepping by default; use
                      OutputDebugString ("ssp off") to disable stepping
 -h, --help           output usage information and exit
 -l, --dll            enable dll profiling.  A chart of relative DLL usage
                      is produced after the run.
 -s, --sub-threads    trace sub-threads too.  Dangerous if you have
                      race conditions.
 -t, --trace-eip      trace every EIP value to a file TRACE.SSP.  This
                      gets big *fast*.
 -v, --verbose        output verbose messages about debug events.
 -V, --version        output version information and exit

Example: ssp 0x401000 0x403000 hello.exe

SSP - The Single Step Profiler

Original Author: DJ Delorie

The SSP is a program that uses the Win32 debug API to run a program one ASM instruction at a time. It records the location of each instruction used, how many times that instruction is used, and all function calls. The results are saved in a format that is usable by the profiling program gprof, although gprof will claim the values are seconds, they really are instruction counts. More on that later.

Because the SSP was originally designed to profile the cygwin DLL, it does not automatically select a block of code to report statistics on. You must specify the range of memory addresses to keep track of manually, but it's not hard to figure out what to specify. Use the "objdump" program to determine the bounds of the target's ".text" section. Let's say we're profiling cygwin1.dll. Make sure you've built it with debug symbols (else gprof won't run) and run objdump like this:

$ objdump -h cygwin1.dll

cygwin1.dll:     file format pei-i386

Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 .text         0007ea00  61001000  61001000  00000400  2**2
                  CONTENTS, ALLOC, LOAD, READONLY, CODE, DATA
  1 .data         00008000  61080000  61080000  0007ee00  2**2
                  CONTENTS, ALLOC, LOAD, DATA
  . . .

The only information we're concerned with are the VMA of the .text section and the VMA of the section after it (sections are usually contiguous; you can also add the Size to the VMA to get the end address). In this case, the VMA is 0x61001000 and the ending address is either 0x61080000 (start of .data method) or 0x0x6107fa00 (VMA+Size method).

There are two basic ways to use SSP - either profiling a whole program, or selectively profiling parts of the program.

To profile a whole program, just run ssp without options. By default, it will step the whole program. Here's a simple example, using the numbers above:

$ ssp 0x61001000 0x61080000 hello.exe

$ gprof -b cygwin1.dll

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
 time   seconds   seconds    calls  ms/call  ms/call  name
 10.02    231.22    72.43       46  1574.57  1574.57  strcspn
  7.95    288.70    57.48      130   442.15   442.15  strncasematch

OK, so now we've got a huge report that took a long time to generate, and we've identified a spot we want to work on optimizing. Let's say it's the time() function. We can use SSP to selectively profile this function by using OutputDebugString() to control SSP from within the program. Here's a sample program:

	#include <windows.h>
	main()
	{
	  time_t t;
	  OutputDebugString("ssp on");
	  time(&t);
	  OutputDebugString("ssp off");
	}

Then, add the -d option to ssp to default to *disabling* profiling. The program will run at full speed until the first OutputDebugString, then step until the second. You can then use gprof (as usual) to see the performance profile for just that portion of the program's execution.

There are many options to ssp. Since step-profiling makes your program run about 1,000 times slower than normal, it's best to understand all the options so that you can narrow down the parts of your program you need to single-step.

-v - verbose. This prints messages about threads starting and stopping, OutputDebugString calls, DLLs loading, etc.

-t and -c - tracing. With -t, *every* step's address is written to the file "trace.ssp". This can be used to help debug functions, since it can trace multiple threads. Clever use of scripts can match addresses with disassembled opcodes if needed. Warning: creates *huge* files, very quickly. -c prints each address to the console, useful for debugging key chunks of assembler. Use addr2line -C -f -s -e foo.exe < trace.ssp > lines.ssp and then perl cvttrace to convert to symbolic traces.

-s - subthreads. Usually, you only need to trace the main thread, but sometimes you need to trace all threads, so this enables that. It's also needed when you want to profile a function that only a subthread calls. However, using OutputDebugString automatically enables profiling on the thread that called it, not the main thread.

-l - dll profiling. Generates a pretty table of how much time was spent in each dll the program used. No sense optimizing a function in your program if most of the time is spent in the DLL. I usually use the -v, -s, and -l options:

$ ssp -v -s -l -d 0x61001000 0x61080000 hello.exe

3.6.14. strace

Usage: strace.exe [OPTIONS] <command-line>
Usage: strace.exe [OPTIONS] -p <pid>
Trace system calls and signals

  -b, --buffer-size=SIZE       set size of output file buffer
  -d, --no-delta               don't display the delta-t microsecond timestamp
  -f, --trace-children         trace child processes (toggle - default true)
  -h, --help                   output usage information and exit
  -m, --mask=MASK              set message filter mask
  -n, --crack-error-numbers    output descriptive text instead of error
                               numbers for Windows errors
  -o, --output=FILENAME        set output file to FILENAME
  -p, --pid=n                  attach to executing program with cygwin pid n
  -S, --flush-period=PERIOD    flush buffered strace output every PERIOD secs
  -t, --timestamp              use an absolute hh:mm:ss timestamp insted of 
                               the default microsecond timestamp.  Implies -d
  -T, --toggle                 toggle tracing in a process already being
                               traced. Requires -p <pid>
  -v, --version                output version information and exit
  -w, --new-window             spawn program under test in a new window

    MASK can be any combination of the following mnemonics and/or hex values
    (0x is optional).  Combine masks with '+' or ',' like so:

                      --mask=wm+system,malloc+0x00800

    Mnemonic Hex     Corresponding Def  Description
    =========================================================================
    all      0x00001 (_STRACE_ALL)      All strace messages.
    flush    0x00002 (_STRACE_FLUSH)    Flush output buffer after each message.
    inherit  0x00004 (_STRACE_INHERIT)  Children inherit mask from parent.
    uhoh     0x00008 (_STRACE_UHOH)     Unusual or weird phenomenon.
    syscall  0x00010 (_STRACE_SYSCALL)  System calls.
    startup  0x00020 (_STRACE_STARTUP)  argc/envp printout at startup.
    debug    0x00040 (_STRACE_DEBUG)    Info to help debugging. 
    paranoid 0x00080 (_STRACE_PARANOID) Paranoid info.
    termios  0x00100 (_STRACE_TERMIOS)  Info for debugging termios stuff.
    select   0x00200 (_STRACE_SELECT)   Info on ugly select internals.
    wm       0x00400 (_STRACE_WM)       Trace Windows msgs (enable _strace_wm).
    sigp     0x00800 (_STRACE_SIGP)     Trace signal and process handling.
    minimal  0x01000 (_STRACE_MINIMAL)  Very minimal strace output.
    exitdump 0x04000 (_STRACE_EXITDUMP) Dump strace cache on exit.
    system   0x08000 (_STRACE_SYSTEM)   Serious error; goes to console and log.
    nomutex  0x10000 (_STRACE_NOMUTEX)  Don't use mutex for synchronization.
    malloc   0x20000 (_STRACE_MALLOC)   Trace malloc calls.
    thread   0x40000 (_STRACE_THREAD)   Thread-locking calls.

The strace program executes a program, and optionally the children of the program, reporting any Cygwin DLL output from the program(s) to stdout, or to a file with the -o option. With the -w option, you can start an strace session in a new window, for example:

$ strace -o tracing_output -w sh -c 'while true; do echo "tracing..."; done' &

Note that strace is a standalone Windows program and so does not rely on the Cygwin DLL itself (you can verify this with cygcheck). As a result it does not understand POSIX pathnames or symlinks. This program is mainly useful for debugging the Cygwin DLL itself.

3.6.15. umount

Usage: umount.exe [OPTION] [<posixpath>]
Unmount filesystems

  -A, --remove-all-mounts       remove all mounts
  -c, --remove-cygdrive-prefix  remove cygdrive prefix
  -h, --help                    output usage information and exit
  -s, --system                  remove system mount (default)
  -S, --remove-system-mounts    remove all system mounts
  -u, --user                    remove user mount
  -U, --remove-user-mounts      remove all user mounts
  -v, --version                 output version information and exit

The umount program removes mounts from the mount table. If you specify a POSIX path that corresponds to a current mount point, umount will remove it from the system registry area. (Administrator priviledges are required). The -u flag may be used to specify removing the mount from the user-specific registry area instead.

The umount utility may also be used to remove all mounts of a particular type. With the extended options it is possible to remove all mounts (-A), all cygdrive automatically-mounted mounts (-c), all mounts in the current user's registry area (-U), or all mounts in the system-wide registry area (-S) (with Administrator privileges).

See Section 3.6.8 for more information on the mount table.

3.7.1. Pathnames

Windows programs do not understand POSIX pathnames, so any arguments that reference the filesystem must be in Windows (or DOS) format or translated. Cygwin provides the cygpath utility for converting between Windows and POSIX paths. A complete description of its options and examples of its usage are in Section 3.6.2, including a shell script for starting Windows Explorer in any directory. The same format works for most Windows programs, for example

notepad.exe "$(cygpath -aw "Desktop/Phone Numbers.txt")"

javac -cp "$(cygpath -pw "$CLASSPATH")" hello.java

3.7.2. Console Programs

Another issue is receiving output from or giving input to the console-based Windows programs. Unfortunately, interacting with Windows console applications is not a simple matter of using a translation utility. Windows console applications and designed to run under command.com or cmd.exe, and some do not deal gracefully with other situations. Cygwin can receive console input only if it is also running in a console (DOS box) since Windows does not provide any way to attach to the backend of the console device. Another traditional Unix input/output method, ptys (pseudo-terminals), are supported by Cygwin but not entirely by Windows. The basic problem is that a Cygwin pty is a pipe and some Windows applications do not like having their input or output redirected to pipes.

To help deal with these issues, Cygwin supports customizable levels of Windows verses Unix compatibility behavior. To be most compatible with Windows programs, use a DOS prompt, running only the occasional Cygwin command or script. Next would be to run bash with the default DOS box. To make Cygwin more Unix compatible in this case, set CYGWIN=tty (see Section 3.5). Alternatively, the optional rxvt package provides a native-Windows version of the popular X11 terminal emulator (it is not necessary to set CYGWIN=tty with rxvt). Using rxvt.exe provides the most Unix-like environment, but expect some compatibility problems with Windows programs.

3.7.3. Cygwin and Windows Networking

Many popular Cygwin packages, such as ncftp, lynx, and wget, require a network connection. Since Cygwin relies on Windows for connectivity, if one of these tools is not working as expected you may need to troubleshoot using Windows tools. The first test is to see if you can reach the URL's host with ping.exe, one of the few utilities included with every Windows version since Windows 95. If you chose to install the inetutils package, you may have both Windows and Cygwin versions of utilities such as ftp and telnet. If you are having problems using one of these programs, see if the alternate one works as expected.

There are a variety of other programs available for specific situations. If your system does not have an always-on network connection, you may be interested in rasdial.exe (or alternatives for Windows 95, 98, and ME) for automating dialup connections. Users who frequently change their network configuration can script these changes with netsh.exe (Windows 2000 and XP). For proxy users, the open source NTLM Authorization Proxy Server or the no-charge Hummingbird SOCKS Proxy may allow you to use Cygwin network programs in your environment.

3.7.4. The cygutils package

The optional cygutils package contains miscellaneous tools that are small enough to not require their own package. It is not included in a default Cygwin install; select it from the Utils category in setup.exe. Several of the cygutils tools are useful for interacting with Windows.

One of the hassles of Unix-Windows interoperability is the different line endings on text files. As mentioned in Section 3.2, Unix tools such as tr can convert between CRLF and LF endings, but cygutils provides several dedicated programs: conv, d2u, dos2unix, u2d, and unix2dos. Use the --help switch for usage information.

3.7.5. Creating shortcuts with cygutils

Another problem area is between Unix-style links, which link one file to another, and Microsoft .lnk files, which provide a shortcut to a file. They seem similar at first glance but, in reality, are fairly different. By default, Cygwin uses a mechanism that creates symbolic links that are compatible with standard Microsoft .lnk files. However, they do not include much of the information that is available in a standard Microsoft shortcut, such as the working directory, an icon, etc. The cygutils package includes a mkshortcut utility for creating standard Microsoft .lnk files.

If Cygwin handled these native shortcuts like any other symlink, you could not archive Microsoft .lnk files into tar archives and keep all the information in them. After unpacking, these shortcuts would have lost all the extra information and would be no different than standard Cygwin symlinks. Therefore these two types of links are treated differently. Unfortunately, this means that the usual Unix way of creating and using symlinks does not work with Windows shortcuts.

3.7.6. Printing with cygutils

There are several options for printing from Cygwin, including the lpr found in cygutils (not to be confused with the native Windows lpr.exe). The easiest way to use cygutils' lpr is to specify a default device name in the PRINTER environment variable. You may also specify a device on the command line with the -d or -P options, which will override the environment variable setting.

A device name may be a UNC path (\\server_name\printer_name), a reserved DOS device name (prn, lpt1), or a local port name that is mapped to a printer share. Note that forward slashes may be used in a UNC path (//server_name/printer_name), which is helpful when using lpr from a shell that uses the backslash as an escape character.

lpr sends raw data to the printer; no formatting is done. Many, but not all, printers accept plain text as input. If your printer supports PostScript, packages such as a2ps and enscript can prepare text files for printing. The ghostscript package also provides some translation from PostScript to various native printer languages. Additionally, a native Windows application for printing PostScript, gsprint, is available from the Ghostscript website.

4.1.1. Console Mode Applications

Use gcc to compile, just like under UNIX. Refer to the GCC User's Guide for information on standard usage and options. Here's a simple example:

C:\> gcc hello.c -o hello.exe
C:\> hello.exe
Hello, World

C:\>

4.1.2. GUI Mode Applications

Cygwin allows you to build programs with full access to the standard Windows 32-bit API, including the GUI functions as defined in any Microsoft or off-the-shelf publication. However, the process of building those applications is slightly different, as you'll be using the GNU tools instead of the Microsoft tools.

For the most part, your sources won't need to change at all. However, you should remove all __export attributes from functions and replace them like this:

int foo (int) __attribute__ ((__dllexport__));

int
foo (int i)

The Makefile is similar to any other UNIX-like Makefile, and like any other Cygwin makefile. The only difference is that you use gcc -mwindows to link your program into a GUI application instead of a command-line application. Here's an example:

myapp.exe : myapp.o myapp.res
	gcc -mwindows myapp.o myapp.res -o $@

myapp.res : myapp.rc resource.h
	windres $< -O coff -o $@

Note the use of windres to compile the Windows resources into a COFF-format .res file. That will include all the bitmaps, icons, and other resources you need, into one handy object file. Normally, if you omitted the "-O coff" it would create a Windows .res format file, but we can only link COFF objects. So, we tell windres to produce a COFF object, but for compatibility with the many examples that assume your linker can handle Windows resource files directly, we maintain the .res naming convention. For more information on windres, consult the Binutils manual.

The following is a simple GUI-mode "Hello, World!" program to help get you started:

/*-------------------------------------------------*/
/* hellogui.c - gui hello world                    */
/* build: gcc -mwindows hellogui.c -o hellogui.exe */
/*-------------------------------------------------*/
#include <windows.h>

char glpszText[1024];

LRESULT CALLBACK WndProc(HWND, UINT, WPARAM, LPARAM);

int APIENTRY WinMain(HINSTANCE hInstance, 
		HINSTANCE hPrevInstance,
		LPSTR lpCmdLine,
		int nCmdShow)
{
	sprintf(glpszText, 
		"Hello World\nGetCommandLine(): [%s]\n"
		"WinMain lpCmdLine: [%s]\n",
		lpCmdLine, GetCommandLine() );

	WNDCLASSEX wcex; 
 
	wcex.cbSize = sizeof(wcex);
	wcex.style = CS_HREDRAW | CS_VREDRAW;
	wcex.lpfnWndProc = WndProc;
	wcex.cbClsExtra = 0;
	wcex.cbWndExtra = 0;
	wcex.hInstance = hInstance;
	wcex.hIcon = LoadIcon(NULL, IDI_APPLICATION);
	wcex.hCursor = LoadCursor(NULL, IDC_ARROW);
	wcex.hbrBackground = (HBRUSH)(COLOR_WINDOW+1);
	wcex.lpszMenuName = NULL;
	wcex.lpszClassName = "HELLO";
	wcex.hIconSm = NULL;

	if (!RegisterClassEx(&wcex))
		return FALSE; 

	HWND hWnd;
	hWnd = CreateWindow("HELLO", "Hello", WS_OVERLAPPEDWINDOW,
		CW_USEDEFAULT, CW_USEDEFAULT, CW_USEDEFAULT, CW_USEDEFAULT, NULL, NULL, hInstance, NULL);

	if (!hWnd)
		return FALSE;

	ShowWindow(hWnd, nCmdShow);
	UpdateWindow(hWnd);

	MSG msg;
	while (GetMessage(&msg, NULL, 0, 0)) 
	{
		TranslateMessage(&msg);
		DispatchMessage(&msg);
	}

	return msg.wParam;
}

LRESULT CALLBACK WndProc(HWND hWnd, UINT message, WPARAM wParam, LPARAM lParam)
{
	PAINTSTRUCT ps;
	HDC hdc;
	
	switch (message) 
	{
		case WM_PAINT:
			hdc = BeginPaint(hWnd, &ps);
			RECT rt;
			GetClientRect(hWnd, &rt);
			DrawText(hdc, glpszText, strlen(glpszText), &rt, DT_TOP | DT_LEFT);
			EndPaint(hWnd, &ps);
			break;
		case WM_DESTROY:
			PostQuitMessage(0);
			break;
		default:
			return DefWindowProc(hWnd, message, wParam, lParam);
	}
	return 0;
}

4.3.1. Building DLLs

This page gives only a few simple examples of gcc's DLL-building capabilities. To begin an exploration of the many additional options, see the gcc documentation and website, currently at http://gcc.gnu.org/

Let's go through a simple example of how to build a dll. For this example, we'll use a single file myprog.c for the program (myprog.exe) and a single file mydll.c for the contents of the dll (mydll.dll).

Fortunately, with the latest gcc and binutils the process for building a dll is now pretty simple. Say you want to build this minimal function in mydll.c:

#include <stdio.h>

int
hello()
{
  printf ("Hello World!\n");
}  

First compile mydll.c to object code:

gcc -c mydll.c

Then, tell gcc that it is building a shared library:

gcc -shared -o mydll.dll mydll.o

That's it! To finish up the example, you can now link to the dll with a simple program:

int 
main ()
{
  hello ();
}  

Then link to your dll with a command like:

gcc -o myprog myprog.ca -L./ -lmydll

However, if you are building a dll as an export library, you will probably want to use the complete syntax:

gcc -shared -o cyg${module}.dll \
    -Wl,--out-implib=lib${module}.dll.a \
    -Wl,--export-all-symbols \
    -Wl,--enable-auto-import \
    -Wl,--whole-archive ${old_lib} \
    -Wl,--no-whole-archive ${dependency_libs}

Where ${module} is the name of your DLL, ${old_lib} are all your object files, bundled together in static libs or single object files and the ${dependency_libs} are import libs you need to link against, e.g '-lpng -lz -L/usr/local/special -lmyspeciallib'.

4.3.2. Linking Against DLLs

If you have an existing DLL already, you need to build a Cygwin-compatible import library. If you have the source to compile the DLL, see Section 4.3.1 for details on having gcc build one for you. If you do not have the source or a supplied working import library, you can get most of the way by creating a .def file with these commands (you might need to do this in bash for the quoting to work correctly):

echo EXPORTS > foo.def
nm foo.dll | grep ' T _' | sed 's/.* T _//' >> foo.def

Note that this will only work if the DLL is not stripped. Otherwise you will get an error message: "No symbols in foo.dll".

Once you have the .def file, you can create an import library from it like this:

dlltool --def foo.def --dllname foo.dll --output-lib foo.a