[PATCH 06/18] mingw: compute the correct size for symlinks in `mingw_lstat()`
- From
- Bill Zissimopoulos via GitGitGadget <gitgitgadget@gmail.com>
- Date
- Dec 17, 2025, 14:08 UTC
- Message-ID
- <4c49a3d9bf57415ea31ac357452b646c75b64d82.1765980535.git.gitgitgadget@gmail.com>
- In-Reply-To
- <pull.2018.git.1765980535.gitgitgadget@gmail.com>
From: Bill Zissimopoulos <billziss@navimatics.com>
POSIX specifies that upon successful return from `lstat()`: "the value of the st_size member shall be set to the length of the pathname contained in the symbolic link not including any terminating null byte".
Git typically doesn't trust the `stat.st_size` member of symlinks (e.g. see `strbuf_readlink()`). Therefore, it is tempting to save on the extra overhead of opening and reading the reparse point merely to calculate the exact size of the link target.
This is, in fact, what Git for Windows did, from May 2015 to May 2020. At least almost: some functions take shortcuts if `st_size` is 0 (e.g. `diff_populate_filespec()`), hence Git for Windows hard-coded the length of all symlinks to MAX_PATH.
This did cause problems, though, specifically in Git repositories that were also accessed by Git for Cygwin or Git for WSL. For example, doing `git reset --hard` using Git for Windows would update the size of symlinks in the index to be MAX_PATH; at a later time Git for Cygwin or Git for WSL would find that symlinks have changed size during `git status` and update the index. And then Git for Windows would think that the index needs to be updated. Even if the symlinks did not, in fact, change. To avoid that, the correct size must be determined.
Signed-off-by: Bill Zissimopoulos <billziss@navimatics.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> --- compat/mingw.c | 114 ++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 104 insertions(+), 10 deletions(-)
diff --git a/compat/mingw.c b/compat/mingw.c index a3a48db581..c7571951dc 100644 --- a/compat/mingw.c +++ b/compat/mingw.c @@ -21,6 +21,7 @@ #define SECURITY_WIN32 #include <sspi.h> #include <wchar.h> +#include <winioctl.h> #include <winternl.h> #define STATUS_DELETE_PENDING ((NTSTATUS) 0xC0000056) @@ -917,10 +918,102 @@ static int has_valid_directory_prefix(wchar_t *wfilename) return 1; } +#ifndef _WINNT_H +/* + * The REPARSE_DATA_BUFFER structure is defined in the Windows DDK (in + * ntifs.h) and in MSYS1's winnt.h (which defines _WINNT_H). So define + * it ourselves if we are on MSYS2 (whose winnt.h defines _WINNT_). + */ +typedef struct _REPARSE_DATA_BUFFER { + DWORD ReparseTag; + WORD ReparseDataLength; + WORD Reserved; +#ifndef _MSC_VER + _ANONYMOUS_UNION +#endif + union { + struct { + WORD SubstituteNameOffset; + WORD SubstituteNameLength; + WORD PrintNameOffset; + WORD PrintNameLength; + ULONG Flags; + WCHAR PathBuffer[1]; + } SymbolicLinkReparseBuffer; + struct { + WORD SubstituteNameOffset; + WORD SubstituteNameLength; + WORD PrintNameOffset; + WORD PrintNameLength; + WCHAR PathBuffer[1]; + } MountPointReparseBuffer; + struct { + BYTE DataBuffer[1]; + } GenericReparseBuffer; + } DUMMYUNIONNAME; +} REPARSE_DATA_BUFFER, *PREPARSE_DATA_BUFFER; +#endif + +static int read_reparse_point(const WCHAR *wpath, BOOL fail_on_unknown_tag, + char *tmpbuf, int *plen, DWORD *ptag) +{ + HANDLE handle; + WCHAR *wbuf; + REPARSE_DATA_BUFFER *b = alloca(MAXIMUM_REPARSE_DATA_BUFFER_SIZE); + DWORD dummy; + + /* read reparse point data */ + handle = CreateFileW(wpath, 0, + FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, + OPEN_EXISTING, + FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OPEN_REPARSE_POINT, NULL); + if (handle == INVALID_HANDLE_VALUE) { + errno = err_win_to_posix(GetLastError()); + return -1; + } + if (!DeviceIoControl(handle, FSCTL_GET_REPARSE_POINT, NULL, 0, b, + MAXIMUM_REPARSE_DATA_BUFFER_SIZE, &dummy, NULL)) { + errno = err_win_to_posix(GetLastError()); + CloseHandle(handle); + return -1; + } + CloseHandle(handle); + + /* get target path for symlinks or mount points (aka 'junctions') */ + switch ((*ptag = b->ReparseTag)) { + case IO_REPARSE_TAG_SYMLINK: + wbuf = (WCHAR*) (((char*) b->SymbolicLinkReparseBuffer.PathBuffer) + + b->SymbolicLinkReparseBuffer.SubstituteNameOffset); + *(WCHAR*) (((char*) wbuf) + + b->SymbolicLinkReparseBuffer.SubstituteNameLength) = 0; + break; + case IO_REPARSE_TAG_MOUNT_POINT: + wbuf = (WCHAR*) (((char*) b->MountPointReparseBuffer.PathBuffer) + + b->MountPointReparseBuffer.SubstituteNameOffset); + *(WCHAR*) (((char*) wbuf) + + b->MountPointReparseBuffer.SubstituteNameLength) = 0; + break; + default: + if (fail_on_unknown_tag) { + errno = EINVAL; + return -1; + } else { + *plen = MAX_PATH; + return 0; + } + } + + if ((*plen = + xwcstoutf(tmpbuf, normalize_ntpath(wbuf), MAX_PATH)) < 0) + return -1; + return 0; +} + int mingw_lstat(const char *file_name, struct stat *buf) { WIN32_FILE_ATTRIBUTE_DATA fdata; - WIN32_FIND_DATAW findbuf = { 0 }; + DWORD reparse_tag = 0; + int link_len = 0; wchar_t wfilename[MAX_PATH]; int wlen = xutftowcs_path(wfilename, file_name); if (wlen < 0) @@ -935,28 +1028,29 @@ int mingw_lstat(const char *file_name, struct stat *buf) } if (GetFileAttributesExW(wfilename, GetFileExInfoStandard, &fdata)) { - /* for reparse points, use FindFirstFile to get the reparse tag */ + /* for reparse points, get the link tag and length */ if (fdata.dwFileAttributes & FILE_ATTRIBUTE_REPARSE_POINT) { - HANDLE handle = FindFirstFileW(wfilename, &findbuf); - if (handle == INVALID_HANDLE_VALUE) - goto error; - FindClose(handle); + char tmpbuf[MAX_PATH]; + + if (read_reparse_point(wfilename, FALSE, tmpbuf, + &link_len, &reparse_tag) < 0) + return -1; } buf->st_ino = 0; buf->st_gid = 0; buf->st_uid = 0; buf->st_nlink = 1; buf->st_mode = file_attr_to_st_mode(fdata.dwFileAttributes, - findbuf.dwReserved0); - buf->st_size = fdata.nFileSizeLow | - (((off_t)fdata.nFileSizeHigh)<<32); + reparse_tag); + buf->st_size = S_ISLNK(buf->st_mode) ? link_len : + fdata.nFileSizeLow | (((off_t) fdata.nFileSizeHigh) << 32); buf->st_dev = buf->st_rdev = 0; /* not used by Git */ filetime_to_timespec(&(fdata.ftLastAccessTime), &(buf->st_atim)); filetime_to_timespec(&(fdata.ftLastWriteTime), &(buf->st_mtim)); filetime_to_timespec(&(fdata.ftCreationTime), &(buf->st_ctim)); return 0; } -error: + switch (GetLastError()) { case ERROR_ACCESS_DENIED: case ERROR_SHARING_VIOLATION:
-- gitgitgadget