| Message ID | 20250313-b4-pks-mingw-lockfile-flake-v1-2-bc5d3e70f516@pks.im (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | compat/mingw: fix EACCESS when opening files with `O_CREAT | O_EXCL` | expand |
Patrick Steinhardt <ps@pks.im> writes: > + /* > + * Internally, `_wopen()` uses the `CreateFile()` API with CREATE_NEW, > + * which may error out with ERROR_ACCESS_DENIED when the file is > + * scheduled for deletion via `DeleteFileW()`. The file essentially > + * exists, so we map this error to ERROR_ALREADY_EXISTS so that callers > + * don't have to special-case this. > + * > + * This fixes issues for example with the lockfile interface when one > + * process has a lock that it is about to commit or release while > + * another process wants to acquire it. > + */ The above may explain how the code gets ERROR_ACCESS_DENIED when there is a pending DeleteFileW() on the file. I however cannot judge if the opposite is also always true, i.e. ERROR_ACCESS_DENIED always mean the file did exist and hasn't gone away and no other reason the code would ever get that error status. Somebody with better understanding on Windows API behaviour hopefully can review it. Thanks. > + if (fd < 0 && create && GetLastError() == ERROR_ACCESS_DENIED) > + errno = EEXIST; > if (fd < 0 && (oflags & O_ACCMODE) != O_RDONLY && errno == EACCES) { > DWORD attrs = GetFileAttributesW(wfilename); > if (attrs != INVALID_FILE_ATTRIBUTES && (attrs & FILE_ATTRIBUTE_DIRECTORY))
Hi Patrick, On Thu, 13 Mar 2025, Patrick Steinhardt wrote: > In our CI systems we can observe that t0610 fails rather frequently. > This testcase races a bunch of git-update-ref(1) processes with one > another which are all trying to update a unique reference, where we > expect that all processes succeed and end up updating the reftable > stack. The error message in this case looks like the following: > > fatal: update_ref failed for ref 'refs/heads/branch-88': reftable: transaction prepare: I/O error I saw this error plenty of times and was wondering whether there would be a way to get more useful information in the error message. After all, I/O errors come in all shapes and forms, and telling the user that _something_ was wrong but forcing them to recreate the issue in a GDB session is an excellent recipe to cause frustration. So I'd like to suggest to improve the user experience substantially by augmenting the rather generic `I/O error` with details as to what operation failed, with what exact error, on what file. > Instrumenting the code with a couple of calls to `BUG()` in relevant > sites where we return `REFTABLE_IO_ERROR` quickly leads one to discover > that this error is caused when calling `flock_acquire()`, which is a > thin wrapper around our lockfile API. Curiously, the error code we get > in such cases is `EACCESS`, indicating that we are not allowed to access > the file. > > The root cause of this is an oddity of `CreateFileW()`, which is what > `_wopen()` uses internally. Quoting its documentation [1]: > > If you call CreateFile on a file that is pending deletion as a > result of a previous call to DeleteFile, the function fails. The > operating system delays file deletion until all handles to the file > are closed. GetLastError returns ERROR_ACCESS_DENIED. > > This behaviour is triggered quite often in the above testcase because > all the processes race with one another trying to acquire the lock for > the "tables.list" file. This is due to how locking works in the reftable > library when compacting a stack: > > 1. Lock the "tables.list" file and reads its contents. > > 2. Decide which tables to compact. > > 3. Lock each of the individual tables that we are about to compact. > > 4. Unlock the "tables.list" file. > > 5. Compact the individual tables into one large table. > > 6. Re-lock the "tables.list" file. > > 7. Write the new list of tables into it. > > 8. Commit the "tables.list" file. > > The important step is (4): we don't commit the file directly by renaming > it into place, but instead we delete the lockfile so that concurrent > processes can continue to append to the reftable stack while we compact > the tables. And because we use `DeleteFileW()` to do so, we may now race > with another process that wants to acquire that lockfile. So if we are > unlucky, we would now see `ERROR_ACCESS_DENIED` instead of the expected > `ERROR_FILE_EXISTS`, which the lockfile subsystem isn't prepared to > handle and thus it will bail out without retrying to acquire the lock. > > In theory, the issue is not limited to the reftable library and can be > triggered by every other user of the lockfile subsystem, as well. My gut > feeling tells me it's rather unlikely to surface elsewhere though. > > Fix the issue by translating the error to `EEXIST`. This makes the > lockfile subsystem handle the error correctly: in case a timeout is set > it will now retry acquiring the lockfile until the timeout has expired. > > With this, t0610 is now always passing on my machine whereas it was > previously failing in around 20-30% of all test runs. It is good that you fixed this issue! However, `ERROR_ACCESS_DENIED` most often means one of two things: - The file in question exists but is opened exclusively by another process (which might be Defender, the anti-malware scanner), or - The current user lacks the permission to create this particular file, i.e. it is really what `EACCES` would mean on Linux. While the first condition clearly can be interpreted as "file exists" in the way this patch wants to do, the latter cannot be. And the patch touches a function that is exclusively used by the `lockfile` machinery, each and every caller of `open(..., ... O_CREAT)` is affected by this change. This has ramifications e.g. when running in a worktree where the user has no write permission (but which they indicated as safe via `safe.directory`). Git would then no longer report correctly whe it cannot write files because the user lacks permission to do that, but would instead claim that the file already exists, when that is not true. Maybe there is a place higher in the stack trace where Git could instead learn to handle `EACCES`? E.g. treat it the same as `EEXIST`, or maybe alternatively make it Windows-specific and introduce a back-off plan? Ciao, Johannes > > [1]: https://learn.microsoft.com/en-us/windows/win32/api/fileapi/nf-fileapi-createfilew > > Signed-off-by: Patrick Steinhardt <ps@pks.im> > --- > compat/mingw.c | 13 +++++++++++++ > 1 file changed, 13 insertions(+) > > diff --git a/compat/mingw.c b/compat/mingw.c > index 101e380c5a3..fb61de759c7 100644 > --- a/compat/mingw.c > +++ b/compat/mingw.c > @@ -644,6 +644,19 @@ int mingw_open (const char *filename, int oflags, ...) > > fd = open_fn(wfilename, oflags, mode); > > + /* > + * Internally, `_wopen()` uses the `CreateFile()` API with CREATE_NEW, > + * which may error out with ERROR_ACCESS_DENIED when the file is > + * scheduled for deletion via `DeleteFileW()`. The file essentially > + * exists, so we map this error to ERROR_ALREADY_EXISTS so that callers > + * don't have to special-case this. > + * > + * This fixes issues for example with the lockfile interface when one > + * process has a lock that it is about to commit or release while > + * another process wants to acquire it. > + */ > + if (fd < 0 && create && GetLastError() == ERROR_ACCESS_DENIED) > + errno = EEXIST; > if (fd < 0 && (oflags & O_ACCMODE) != O_RDONLY && errno == EACCES) { > DWORD attrs = GetFileAttributesW(wfilename); > if (attrs != INVALID_FILE_ATTRIBUTES && (attrs & FILE_ATTRIBUTE_DIRECTORY)) > > -- > 2.49.0.rc2.394.gf6994c5077.dirty > > >
On Sun, Mar 16, 2025 at 01:01:25AM +0100, Johannes Schindelin wrote: > On Thu, 13 Mar 2025, Patrick Steinhardt wrote: > > In our CI systems we can observe that t0610 fails rather frequently. > > This testcase races a bunch of git-update-ref(1) processes with one > > another which are all trying to update a unique reference, where we > > expect that all processes succeed and end up updating the reftable > > stack. The error message in this case looks like the following: > > > > fatal: update_ref failed for ref 'refs/heads/branch-88': reftable: transaction prepare: I/O error > > I saw this error plenty of times and was wondering whether there would be > a way to get more useful information in the error message. > > After all, I/O errors come in all shapes and forms, and telling the user > that _something_ was wrong but forcing them to recreate the issue in a GDB > session is an excellent recipe to cause frustration. > > So I'd like to suggest to improve the user experience substantially by > augmenting the rather generic `I/O error` with details as to what > operation failed, with what exact error, on what file. Agreed, the error handling isn't great. The very least we should be doing is to print `errno`, but even that I consider to be suboptimal. Ideally we'd have structured error handling that allows us to return richer errors to the caller, but that is a much bigger undertaking. [snip] > > Fix the issue by translating the error to `EEXIST`. This makes the > > lockfile subsystem handle the error correctly: in case a timeout is set > > it will now retry acquiring the lockfile until the timeout has expired. > > > > With this, t0610 is now always passing on my machine whereas it was > > previously failing in around 20-30% of all test runs. > > It is good that you fixed this issue! > > However, `ERROR_ACCESS_DENIED` most often means one of two things: > > - The file in question exists but is opened exclusively by another process > (which might be Defender, the anti-malware scanner), or > > - The current user lacks the permission to create this particular file, > i.e. it is really what `EACCES` would mean on Linux. > > While the first condition clearly can be interpreted as "file exists" in > the way this patch wants to do, the latter cannot be. And the patch > touches a function that is exclusively used by the `lockfile` machinery, > each and every caller of `open(..., ... O_CREAT)` is affected by this > change. I feared as much. I was hoping that the second case would cause a different error equivalent to EPERM, and the documentation didn't really say anything about this. > This has ramifications e.g. when running in a worktree where the user has > no write permission (but which they indicated as safe via > `safe.directory`). Git would then no longer report correctly whe it cannot > write files because the user lacks permission to do that, but would > instead claim that the file already exists, when that is not true. > > Maybe there is a place higher in the stack trace where Git could instead > learn to handle `EACCES`? E.g. treat it the same as `EEXIST`, or maybe > alternatively make it Windows-specific and introduce a back-off plan? The place that would need to learn about it is the lockfile subsystem. But we basically have the same issue here that we cannot know why we got EACCESS in the first place. So retrying may or may not be the correct thing to do in this context, same as in `mingw_open()`. While implementing the workaround I wondered whether we are able to get clearer error messages if we were able to verify a few additional data points: - If creating the file fails with ERROR_ACCESS_DENIED we could check whether the parent directory is accessible to us, and if it is then we can assume that the error is due to an existing file. But that falls apart rather quickly when thinking about edge cases, like an unwritable file in a writable directory. - We could stat the file in question to check whether it exists. But given that our case only happens when we have lost a race it may be unwise to build on top of an already-racy mechanism. All of these feel hacky, so... I don't have a good idea for how to fix this. It is unfortunate that `CreateFileW()` throws these two errors into the same bag and doesn't give us any hints which of both errors has happened. Patrick
diff --git a/compat/mingw.c b/compat/mingw.c index 101e380c5a3..fb61de759c7 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -644,6 +644,19 @@ int mingw_open (const char *filename, int oflags, ...) fd = open_fn(wfilename, oflags, mode); + /* + * Internally, `_wopen()` uses the `CreateFile()` API with CREATE_NEW, + * which may error out with ERROR_ACCESS_DENIED when the file is + * scheduled for deletion via `DeleteFileW()`. The file essentially + * exists, so we map this error to ERROR_ALREADY_EXISTS so that callers + * don't have to special-case this. + * + * This fixes issues for example with the lockfile interface when one + * process has a lock that it is about to commit or release while + * another process wants to acquire it. + */ + if (fd < 0 && create && GetLastError() == ERROR_ACCESS_DENIED) + errno = EEXIST; if (fd < 0 && (oflags & O_ACCMODE) != O_RDONLY && errno == EACCES) { DWORD attrs = GetFileAttributesW(wfilename); if (attrs != INVALID_FILE_ATTRIBUTES && (attrs & FILE_ATTRIBUTE_DIRECTORY))
In our CI systems we can observe that t0610 fails rather frequently. This testcase races a bunch of git-update-ref(1) processes with one another which are all trying to update a unique reference, where we expect that all processes succeed and end up updating the reftable stack. The error message in this case looks like the following: fatal: update_ref failed for ref 'refs/heads/branch-88': reftable: transaction prepare: I/O error Instrumenting the code with a couple of calls to `BUG()` in relevant sites where we return `REFTABLE_IO_ERROR` quickly leads one to discover that this error is caused when calling `flock_acquire()`, which is a thin wrapper around our lockfile API. Curiously, the error code we get in such cases is `EACCESS`, indicating that we are not allowed to access the file. The root cause of this is an oddity of `CreateFileW()`, which is what `_wopen()` uses internally. Quoting its documentation [1]: If you call CreateFile on a file that is pending deletion as a result of a previous call to DeleteFile, the function fails. The operating system delays file deletion until all handles to the file are closed. GetLastError returns ERROR_ACCESS_DENIED. This behaviour is triggered quite often in the above testcase because all the processes race with one another trying to acquire the lock for the "tables.list" file. This is due to how locking works in the reftable library when compacting a stack: 1. Lock the "tables.list" file and reads its contents. 2. Decide which tables to compact. 3. Lock each of the individual tables that we are about to compact. 4. Unlock the "tables.list" file. 5. Compact the individual tables into one large table. 6. Re-lock the "tables.list" file. 7. Write the new list of tables into it. 8. Commit the "tables.list" file. The important step is (4): we don't commit the file directly by renaming it into place, but instead we delete the lockfile so that concurrent processes can continue to append to the reftable stack while we compact the tables. And because we use `DeleteFileW()` to do so, we may now race with another process that wants to acquire that lockfile. So if we are unlucky, we would now see `ERROR_ACCESS_DENIED` instead of the expected `ERROR_FILE_EXISTS`, which the lockfile subsystem isn't prepared to handle and thus it will bail out without retrying to acquire the lock. In theory, the issue is not limited to the reftable library and can be triggered by every other user of the lockfile subsystem, as well. My gut feeling tells me it's rather unlikely to surface elsewhere though. Fix the issue by translating the error to `EEXIST`. This makes the lockfile subsystem handle the error correctly: in case a timeout is set it will now retry acquiring the lockfile until the timeout has expired. With this, t0610 is now always passing on my machine whereas it was previously failing in around 20-30% of all test runs. [1]: https://learn.microsoft.com/en-us/windows/win32/api/fileapi/nf-fileapi-createfilew Signed-off-by: Patrick Steinhardt <ps@pks.im> --- compat/mingw.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)