block: Ignore close() failure in get_tmp_filename()
PATCH v6 1/2 block: Ignore close() failure in get_tmp_filename()
The temporary file has been created and is ready for use. Checking return value of close() does not seem useful. The file descriptor is almost certainly closed; see close(2) under "Dealing with error returns from close()".
Let's simply ignore close() failure here.
---
(no changes since v5)
Changes in v5:
- new patch: "block: Ignore close() failure in get_tmp_filename()"
block.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-)
block.c +1 -4
@@ -886,10 +886,7 @@ int get_tmp_filename(char *filename, int size)if (fd <0 ) {return -errno; } -if (close (fd) !=0 ) { -unlink (filename); -return -errno; - } +close (fd);return 0 ;#endif }
Signature · 1 line
2.25.1
PATCH v6 2/2 block: Refactor get_tmp_filename()
At present there are two callers of get_tmp_filename() and they are inconsistent.
One does:
/* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
char *tmp_filename = g_malloc0(PATH_MAX + 1);
...
ret = get_tmp_filename(tmp_filename, PATH_MAX + 1);while the other does:
s->qcow_filename = g_malloc(PATH_MAX);
ret = get_tmp_filename(s->qcow_filename, PATH_MAX);As we can see different 'size' arguments are passed. There are also platform specific implementations inside the function, and the use of snprintf is really undesirable.
The function name is also misleading. It creates a temporary file, not just a filename.
Refactor this routine by changing its name and signature to:
char *create_tmp_file(Error **errp)
and use g_get_tmp_dir() / g_mkstemp() for a consistent implementation.
While we are here, add some comments to mention that /var/tmp is preferred over /tmp on non-win32 hosts.
---
Changes in v6:
- use g_mkstemp() and stick to use /var/tmp for non-win32 hosts
Changes in v5:
- minor change in the commit message
- add some notes in the function comment block
- add g_autofree for tmp_filename
Changes in v4:
- Rename the function to create_tmp_file() and take "Error **errp" as a parameter, so that callers can pass errp all the way down to this routine.
- Commit message updated to reflect the latest change
Changes in v3:
- Do not use errno directly, instead still let get_tmp_filename() return a negative number to indicate error
Changes in v2:
- Use g_autofree and g_steal_pointer
include/block/block_int-common.h | 2 +- block.c | 56 +++++++++++++++++--------------- block/vvfat.c | 7 ++-- 3 files changed, 34 insertions(+), 31 deletions(-)
include/block/block_int-common.h +1 -1
@@ -1230,7 +1230,7 @@ static inline BlockDriverState *child_bs(BdrvChild *child) }int bdrv_check_request (int64_t offset,int64_t bytes,Error **errp); -int get_tmp_filename (char *filename,int size); +char *create_tmp_file (Error **errp);void bdrv_parse_filename_strip_prefix (const char *filename,const char *prefix,QDict *options);
block.c +30 -26
89 lines not on this page · raw message
block/vvfat.c +3 -4
@@ -3146,10 +3146,9 @@ static int enable_write_target(BlockDriverState *bs, Error **errp)array_init (&(s->commits),sizeof (commit_t)); - s->qcow_filename =g_malloc (PATH_MAX ); - ret =get_tmp_filename (s->qcow_filename,PATH_MAX ); -if (ret <0 ) { -error_setg_errno (errp, -ret,"can't create temporary file" ); + s->qcow_filename =create_tmp_file (errp); +if (!s->qcow_filename) { + ret = -ENOENT ; goto err; }
Signature · 1 line
2.25.1
Re: [PATCH v6 2/2] block: Refactor get_tmp_filename()
Quoted 53 lines · Bin Meng
At present there are two callers of
get_tmp_filename()and they are inconsistent.One does:
/* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */ char *tmp_filename = g_malloc0(PATH_MAX + 1); ... ret = get_tmp_filename(tmp_filename, PATH_MAX + 1);while the other does:
s->qcow_filename = g_malloc(PATH_MAX); ret = get_tmp_filename(s->qcow_filename, PATH_MAX);As we can see different 'size' arguments are passed. There are also platform specific implementations inside the function, and the use of snprintf is really undesirable.
The function name is also misleading. It creates a temporary file, not just a filename.
Refactor this routine by changing its name and signature to:
char *create_tmp_file(Error **errp)
and use
g_get_tmp_dir()/g_mkstemp()for a consistent implementation.While we are here, add some comments to mention that
/var/tmpis preferred over /tmp on non-win32 hosts.Signed-off-by Bin Meng <[email protected]>---
Changes in v6:
- useg_mkstemp()and stick to use/var/tmpfor non-win32 hostsChanges in v5:
- minor change in the commit message
- add some notes in the function comment block
- addg_autofreefortmp_filenameChanges in v4:
- Rename the function tocreate_tmp_file()and take "Error **errp" as a parameter, so that callers can pass errp all the way down to this routine.
- Commit message updated to reflect the latest changeChanges in v3:
- Do not use errno directly, instead still letget_tmp_filename()return a negative number to indicate error
Quoted 8 lines
Changes in v2:
- Useg_autofreeandg_steal_pointerinclude/block/block_int-common.h | 2 +- block.c | 56 +++++++++++++++++--------------- block/vvfat.c | 7 ++-- 3 files changed, 34 insertions(+), 31 deletions(-)
Any comments?
Re: [PATCH v6 2/2] block: Refactor get_tmp_filename()
In reply to Bin MengAm 10.10.2022 um 06:04 hat Bin Meng geschrieben:
Quoted 108 lines
At present there are two callers of
get_tmp_filename()and they are inconsistent.One does:
/* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */ char *tmp_filename = g_malloc0(PATH_MAX + 1); ... ret = get_tmp_filename(tmp_filename, PATH_MAX + 1);while the other does:
s->qcow_filename = g_malloc(PATH_MAX); ret = get_tmp_filename(s->qcow_filename, PATH_MAX);As we can see different 'size' arguments are passed. There are also platform specific implementations inside the function, and the use of snprintf is really undesirable.
The function name is also misleading. It creates a temporary file, not just a filename.
Refactor this routine by changing its name and signature to:
char *create_tmp_file(Error **errp)
and use
g_get_tmp_dir()/g_mkstemp()for a consistent implementation.While we are here, add some comments to mention that
/var/tmpis preferred over /tmp on non-win32 hosts.Signed-off-by Bin Meng <[email protected]>---
Changes in v6:
- useg_mkstemp()and stick to use/var/tmpfor non-win32 hostsChanges in v5:
- minor change in the commit message
- add some notes in the function comment block
- addg_autofreefortmp_filenameChanges in v4:
- Rename the function tocreate_tmp_file()and take "Error **errp" as a parameter, so that callers can pass errp all the way down to this routine.
- Commit message updated to reflect the latest changeChanges in v3:
- Do not use errno directly, instead still letget_tmp_filename()return a negative number to indicate errorChanges in v2:
- Useg_autofreeandg_steal_pointerinclude/block/block_int-common.h | 2 +- block.c | 56 +++++++++++++++++--------------- block/vvfat.c | 7 ++-- 3 files changed, 34 insertions(+), 31 deletions(-)include/block/block_int-common.h +1
-1@@ -1230,7 +1230,7 @@ static inline BlockDriverState *child_bs(BdrvChild *child) }int bdrv_check_request (int64_t offset,int64_t bytes,Error **errp); -int get_tmp_filename (char *filename,int size); +char *create_tmp_file (Error **errp);void bdrv_parse_filename_strip_prefix (const char *filename,const char *prefix,QDict *options);block.c +10
-13@@ -860,35 +860,42 @@ int bdrv_probe_geometry(BlockDriverState *bs, HDGeometry *geo)/* * Create a uniquely-named empty temporary file. -* Return 0 upon success, otherwise a negative errno value. +* Return the actual file name used upon success, otherwise NULL. +* This string should be freed with g_free() when not needed any longer. +* +* Note: creating a temporary file for the caller to (re)open is +* inherently racy. Use g_file_open_tmp() instead whenever practical. */ -int get_tmp_filename (char *filename,int size) +char *create_tmp_file (Error **errp) { -#ifdef _WIN32 -char temp_dir[MAX_PATH ]; -/* GetTempFileName requires that its output buffer (4th param) -have length MAX_PATH or greater. */ -assert (size >=MAX_PATH ); -return (GetTempPath (MAX_PATH , temp_dir) - &&GetTempFileName (temp_dir,"qem" ,0 , filename) - ?0 : -GetLastError ()); -#else int fd;const char *tmpdir; - tmpdir =getenv ("TMPDIR" ); -if (!tmpdir) { + g_autofree char *filename = NULL; + + tmpdir =g_get_tmp_dir (); +#ifndef _WIN32
block.c int bdrv_probe_geometry(BlockDriverState *bs, HDGeometry *geo) · line 886
Quoted 10 lines
block.c +8
-0+/* +* See commit 69bef79 ("block: use /var/tmp instead of /tmp for -snapshot") +* +* This function is used to create temporary disk images (like -snapshot), +* so the files can become very large. /tmp is often a tmpfs where as +* /var/tmp is usually on a disk, so more appropriate for disk images. +*/ +if (!g_strcmp0 (tmpdir,"/tmp" )) { tmpdir ="/var/tmp" ; }
This is still a slight change in behaviour: If the TMPDIR environment variable was explicit set to /tmp, QEMU would store the image file in /tmp before this patch. After the patch, it will use /var/tmp even in this case.
I suppose this is a tolerable change.
Kevin
Am 10.10.2022 um 06:04 hat Bin Meng geschrieben:
Quoted 10 lines
The temporary file has been created and is ready for use. Checking return value of
close()does not seem useful. The file descriptor is almost certainly closed; see close(2) under "Dealing with error returns fromclose()".Let's simply ignore
close()failure here.Suggested-by Markus Armbruster <[email protected]>Signed-off-by Bin Meng <[email protected]>Reviewed-by Markus Armbruster <[email protected]>
Thanks, series applied to the block branch.
Kevin