libxfs: stop overriding MAP_SYNC in publicly exported header files
RFC PATCH 1/1 libxfs: stop overriding MAP_SYNC in publicly exported header files
Can one of you please apply this patch and see if it'll build in musl on mips, please? Sorry it's taken so long to address this. :/
--D
---
From: Darrick J. Wong <[email protected]>
Florian Fainelli most recently reported that xfsprogs doesn't build with musl on mips:
"MIPS platforms building with recent kernel headers and the musl-libc toolchain will expose the following build failure:
mmap.c: In function 'mmap_f':
mmap.c:196:12: error: 'MAP_SYNC' undeclared (first use in this function); did you mean 'MS_SYNC'?
196 | flags = MAP_SYNC | MAP_SHARED_VALIDATE;
| ^~~~~~~~
| MS_SYNCmmap.c:196:12: note: each undeclared identifier is reported only once for each function it appears in make[4]: *** [../include/buildrules:81: mmap.o] Error 1"
At first glance, the build failure here is caused by the fact that:
1. The configure script doesn't detect MAP_SYNC support
2. The build system doesn't set HAVE_MAP_SYNC
2. io/mmap.c includes input.h -> projects.h -> xfs.h and later sys/mman.h
3. include/linux.h #define's MAP_SYNC to 0 if HAVE_MAP_SYNC is not set
4. musl's sys/mman.h #undef MAP_SYNC on platforms that don't support it
5. io/mmap.c tries to use MAP_SYNC, not realizing that libc undefined it
Normally, xfs_io only exports functionality that is defined by the libc and/or kernel headers on the build system. We often make exceptions for new functionality so that we have a way to test them before the header file packages catch up, hence this '#ifndef HAVE_FOO #define FOO' paradigm.
MAP_SYNC is a gross and horribly broken example of this. These support crutches are supposed to be *private* to xfsprogs for benefit of early testing, but they were instead added to include/linux.h, which we provide to user programs in the xfslibs-dev package. IOWs, we've been #defining MAP_SYNC to zero for unsuspecting programs.
Worst yet, gcc 11.3 doesn't even warn about overriding a #define to 0:
#include <stdio.h>
#include <sys/mman.h>
#ifdef STUPID
# include <xfs/xfs.h> #endif
int main(int argc, char *argv[]) {
printf("MAP_SYNC 0x%x\n", MAP_SYNC);
}$ gcc -o a a.c -Wall $ ./a MAP_SYNC 0x80000 $ gcc -DSTUPID -o a a.c -Wall $ ./a MAP_SYNC 0x0
Four years have gone by since the introduction of MAP_SYNC, so let's get rid of the override code entirely -- any platform that supports MAP_SYNC has had plenty of chances to ensure their header files have the right bits. While we're at it, fix AC_HAVE_MAP_SYNC to look for MAP_SYNC in the same header file that the one user (io/mmap.c) uses -- sys/mman.h.
Annoyingly, I had to test this by hand because the sole fstest that exercises MAP_SYNC (generic/470) requires dm-logwrites and dm-thinp, neither of which support fsdax on current kernels.
--- include/linux.h | 8 -------- io/io.h | 2 +- io/mmap.c | 25 +++++++++++++------------ m4/package_libcdev.m4 | 3 +-- 4 files changed, 15 insertions(+), 23 deletions(-)
include/linux.h +0 -8
@@ -251,14 +251,6 @@ struct fsxattr {#define FS_XFLAG_COWEXTSIZE 0x00010000/* CoW extent size allocator hint */ #endif -#ifndef HAVE_MAP_SYNC -#define MAP_SYNC 0 -#define MAP_SHARED_VALIDATE 0 -#else -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> -#endif /* HAVE_MAP_SYNC */ -/* * Reminder: anything added to this file will be compiled into downstream * userspace projects!
io/io.h +1 -1
@@ -58,7 +58,7 @@ typedef struct mmap_region {size_t length;/* length of mapping */ off64_t offset;/* start offset into backing file */ int prot;/* protection mode of the mapping */ -bool map_sync;/* is this a MAP_SYNC mapping? */ +int flags;/* MAP_* flags passed to mmap() */ char *name;/* name of backing file */ } mmap_region_t;
io/mmap.c +13 -12
@@ -46,8 +46,11 @@ print_mapping(for (i =0 , p = pflags; p->prot !=PROT_NONE ; i++, p++) buffer[i] = (map->prot & p->prot) ? p->mode :'-' ; -if (map->map_sync) +#ifdef HAVE_MAP_SYNC +if ((map->flags & (MAP_SYNC |MAP_SHARED_VALIDATE )) == + (MAP_SYNC |MAP_SHARED_VALIDATE ))sprintf (&buffer[i]," S" ); +#endif printf("%c%03d%c 0x%lx - 0x%lx %s %14s (%lld : %ld)\n", braces? '[' : ' ', index, braces? ']' : ' ',
@@ -139,7 +142,9 @@ mmap_help(void) " -r -- map with PROT_READ protection\n" " -w -- map with PROT_WRITE protection\n" " -x -- map with PROT_EXEC protection\n" +#ifdef HAVE_MAP_SYNC " -S -- map with MAP_SYNC and MAP_SHARED_VALIDATE flags\n" +#endif" -s <size> -- first do mmap(size)/munmap(size), try to reserve some free space\n" " If no protection mode is specified, all are used by default.\n" "\n"));
@@ -193,18 +198,14 @@ mmap_f( prot |=PROT_EXEC ;break ;case 'S' : +#ifdef HAVE_MAP_SYNC flags =MAP_SYNC |MAP_SHARED_VALIDATE ; - -/* -* If MAP_SYNC and MAP_SHARED_VALIDATE aren't defined -* in the system headers we will have defined them -* both as 0. -*/ -if (!flags) { -printf ("MAP_SYNC not supported\n" ); -return 0 ; - }break ; +#else +printf ("MAP_SYNC not supported\n" ); + exitcode =1 ; +return command_usage (&mmap_cmd); +#endif case 's' : length2 =cvtnum (blocksize, sectsize, optarg);break ;
@@ -281,7 +282,7 @@ mmap_f( mapping->offset = offset; mapping->name = filename; mapping->prot = prot; - mapping->map_sync = (flags == (MAP_SYNC |MAP_SHARED_VALIDATE )); + mapping->flags = flags;return 0 ; }
m4/package_libcdev.m4 +1 -2
@@ -387,8 +387,7 @@ AC_DEFUN([AC_HAVE_MAP_SYNC], [ AC_MSG_CHECKING([for MAP_SYNC]) AC_COMPILE_IFELSE( [ AC_LANG_PROGRAM([[ -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> +#include <sys/mman.h> ]], [[ int flags = MAP_SYNC | MAP_SHARED_VALIDATE; ]])
Quoted 68 lines · Darrick J. Wong
Can one of you please apply this patch and see if it'll build in musl on mips, please? Sorry it's taken so long to address this. :/
--D
---
From: Darrick J. Wong <[email protected]>Florian Fainelli most recently reported that xfsprogs doesn't build with musl on mips:
"MIPS platforms building with recent kernel headers and the musl-libc toolchain will expose the following build failure:
mmap.c: In function 'mmap_f':
mmap.c:196:12: error: 'MAP_SYNC' undeclared (first use in this function); did you mean 'MS_SYNC'?196 | flags = MAP_SYNC | MAP_SHARED_VALIDATE; | ^~~~~~~~ | MS_SYNCmmap.c:196:12: note: each undeclared identifier is reported only once for each function it appears in make[4]: *** [../include/buildrules:81: mmap.o] Error 1"
At first glance, the build failure here is caused by the fact that:
1. The configure script doesn't detect
MAP_SYNCsupport
2. The build system doesn't setHAVE_MAP_SYNC
2.io/mmap.cincludesinput.h->projects.h->xfs.hand latersys/mman.h
3.include/linux.h#define'sMAP_SYNCto 0 ifHAVE_MAP_SYNCis not set
4. musl'ssys/mman.h#undefMAP_SYNCon platforms that don't support it
5.io/mmap.ctries to useMAP_SYNC, not realizing that libc undefined itNormally,
xfs_ioonly exports functionality that is defined by the libc and/or kernel headers on the build system. We often make exceptions for new functionality so that we have a way to test them before the header file packages catch up, hence this '#ifndefHAVE_FOO#define FOO' paradigm.
MAP_SYNCis a gross and horribly broken example of this. These support crutches are supposed to be *private* to xfsprogs for benefit of early testing, but they were instead added toinclude/linux.h, which we provide to user programs in the xfslibs-dev package. IOWs, we've been #definingMAP_SYNCto zero for unsuspecting programs.Worst yet, gcc 11.3 doesn't even warn about overriding a #define to 0:
#include <
stdio.h>
#include <sys/mman.h>
#ifdef STUPID# include <xfs/xfs.h> #endifint main(int argc, char *argv[]) { printf("MAP_SYNC 0x%x\n", MAP_SYNC); }$ gcc -o a a.c -Wall $ ./a MAP_SYNC 0x80000 $ gcc -DSTUPID -o a a.c -Wall $ ./a MAP_SYNC 0x0Four years have gone by since the introduction of
MAP_SYNC, so let's get rid of the override code entirely -- any platform that supportsMAP_SYNChas had plenty of chances to ensure their header files have the right bits. While we're at it, fixAC_HAVE_MAP_SYNCto look forMAP_SYNCin the same header file that the one user (io/mmap.c) uses --sys/mman.h.
Quoted 8 lines
Annoyingly, I had to test this by hand because the sole fstest that exercises
MAP_SYNC(generic/470) requires dm-logwrites and dm-thinp, neither of which support fsdax on current kernels.Reported-by [email protected]Reported-by Fabrice Fontaine <[email protected]>Reported-by Florian Fainelli <[email protected]>Signed-off-by Darrick J. Wong <[email protected]>
Thanks!
Signature · 1 line
Florian
Quoted 181 lines · Darrick J. Wong
Can one of you please apply this patch and see if it'll build in musl on mips, please? Sorry it's taken so long to address this. :/
--D
---
From: Darrick J. Wong <[email protected]>Florian Fainelli most recently reported that xfsprogs doesn't build with musl on mips:
"MIPS platforms building with recent kernel headers and the musl-libc toolchain will expose the following build failure:
mmap.c: In function 'mmap_f':
mmap.c:196:12: error: 'MAP_SYNC' undeclared (first use in this function); did you mean 'MS_SYNC'?196 | flags = MAP_SYNC | MAP_SHARED_VALIDATE; | ^~~~~~~~ | MS_SYNCmmap.c:196:12: note: each undeclared identifier is reported only once for each function it appears in make[4]: *** [../include/buildrules:81: mmap.o] Error 1"
At first glance, the build failure here is caused by the fact that:
1. The configure script doesn't detect
MAP_SYNCsupport
2. The build system doesn't setHAVE_MAP_SYNC
2.io/mmap.cincludesinput.h->projects.h->xfs.hand latersys/mman.h
3.include/linux.h#define'sMAP_SYNCto 0 ifHAVE_MAP_SYNCis not set
4. musl'ssys/mman.h#undefMAP_SYNCon platforms that don't support it
5.io/mmap.ctries to useMAP_SYNC, not realizing that libc undefined itNormally,
xfs_ioonly exports functionality that is defined by the libc and/or kernel headers on the build system. We often make exceptions for new functionality so that we have a way to test them before the header file packages catch up, hence this '#ifndefHAVE_FOO#define FOO' paradigm.
MAP_SYNCis a gross and horribly broken example of this. These support crutches are supposed to be *private* to xfsprogs for benefit of early testing, but they were instead added toinclude/linux.h, which we provide to user programs in the xfslibs-dev package. IOWs, we've been #definingMAP_SYNCto zero for unsuspecting programs.Worst yet, gcc 11.3 doesn't even warn about overriding a #define to 0:
#include <
stdio.h>
#include <sys/mman.h>
#ifdef STUPID# include <xfs/xfs.h> #endifint main(int argc, char *argv[]) { printf("MAP_SYNC 0x%x\n", MAP_SYNC); }$ gcc -o a a.c -Wall $ ./a MAP_SYNC 0x80000 $ gcc -DSTUPID -o a a.c -Wall $ ./a MAP_SYNC 0x0Four years have gone by since the introduction of
MAP_SYNC, so let's get rid of the override code entirely -- any platform that supportsMAP_SYNChas had plenty of chances to ensure their header files have the right bits. While we're at it, fixAC_HAVE_MAP_SYNCto look forMAP_SYNCin the same header file that the one user (io/mmap.c) uses --sys/mman.h.Annoyingly, I had to test this by hand because the sole fstest that exercises
MAP_SYNC(generic/470) requires dm-logwrites and dm-thinp, neither of which support fsdax on current kernels.Reported-by [email protected]Reported-by Fabrice Fontaine <[email protected]>Reported-by Florian Fainelli <[email protected]>Signed-off-by Darrick J. Wong <[email protected]>--- include/linux.h | 8 -------- io/io.h | 2 +- io/mmap.c | 25 +++++++++++++------------ m4/package_libcdev.m4 | 3 +-- 4 files changed, 15 insertions(+), 23 deletions(-)include/linux.h +0
-8@@ -251,14 +251,6 @@ struct fsxattr {#define FS_XFLAG_COWEXTSIZE 0x00010000/* CoW extent size allocator hint */ #endif -#ifndef HAVE_MAP_SYNC -#define MAP_SYNC 0 -#define MAP_SHARED_VALIDATE 0 -#else -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> -#endif /* HAVE_MAP_SYNC */ -/* * Reminder: anything added to this file will be compiled into downstream * userspace projects! io/io.h +1
-1@@ -58,7 +58,7 @@ typedef struct mmap_region {size_t length;/* length of mapping */ off64_t offset;/* start offset into backing file */ int prot;/* protection mode of the mapping */ -bool map_sync;/* is this a MAP_SYNC mapping? */ +int flags;/* MAP_* flags passed to mmap() */ char *name;/* name of backing file */ } mmap_region_t;io/mmap.c +13
-12@@ -46,8 +46,11 @@ print_mapping(for (i =0 , p = pflags; p->prot !=PROT_NONE ; i++, p++) buffer[i] = (map->prot & p->prot) ? p->mode :'-' ; -if (map->map_sync) +#ifdef HAVE_MAP_SYNC +if ((map->flags & (MAP_SYNC |MAP_SHARED_VALIDATE )) == + (MAP_SYNC |MAP_SHARED_VALIDATE ))sprintf (&buffer[i]," S" ); +#endif printf("%c%03d%c 0x%lx - 0x%lx %s %14s (%lld : %ld)\n", braces? '[' : ' ', index, braces? ']' : ' ',@@ -139,7 +142,9 @@ mmap_help(void) " -r -- map with PROT_READ protection\n" " -w -- map with PROT_WRITE protection\n" " -x -- map with PROT_EXEC protection\n" +#ifdef HAVE_MAP_SYNC " -S -- map with MAP_SYNC and MAP_SHARED_VALIDATE flags\n" +#endif" -s <size> -- first do mmap(size)/munmap(size), try to reserve some free space\n" " If no protection mode is specified, all are used by default.\n" "\n"));@@ -193,18 +198,14 @@ mmap_f( prot |=PROT_EXEC ;break ;case 'S' : +#ifdef HAVE_MAP_SYNC flags =MAP_SYNC |MAP_SHARED_VALIDATE ; - -/* -* If MAP_SYNC and MAP_SHARED_VALIDATE aren't defined -* in the system headers we will have defined them -* both as 0. -*/ -if (!flags) { -printf ("MAP_SYNC not supported\n" ); -return 0 ; - }break ; +#else +printf ("MAP_SYNC not supported\n" ); + exitcode =1 ; +return command_usage (&mmap_cmd); +#endif case 's' : length2 =cvtnum (blocksize, sectsize, optarg);break ;@@ -281,7 +282,7 @@ mmap_f( mapping->offset = offset; mapping->name = filename; mapping->prot = prot; - mapping->map_sync = (flags == (MAP_SYNC |MAP_SHARED_VALIDATE )); + mapping->flags = flags;return 0 ; }m4/package_libcdev.m4 +0
-0
m4/package_libcdev.m4 AC_DEFUN([AC_HAVE_MAP_SYNC], · line 393
Quoted 9 lines
m4/package_libcdev.m4 +1
-2@@ -387,8 +387,7 @@ AC_DEFUN([AC_HAVE_MAP_SYNC], [ AC_MSG_CHECKING([for MAP_SYNC]) AC_COMPILE_IFELSE( [ AC_LANG_PROGRAM([[ -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> +#include <sys/mman.h> ]], [[ int flags = MAP_SYNC | MAP_SHARED_VALIDATE; ]])
Patch looks good, if you plan to make it into a non-RFC patch, feel free to include:
Cheers.
Signature · 1 line
Carlos Maiolino
Quoted 195 lines · Carlos Maiolino
Quoted 181 lines · Darrick J. Wong
Can one of you please apply this patch and see if it'll build in musl on mips, please? Sorry it's taken so long to address this. :/
--D
---
From: Darrick J. Wong <[email protected]>Florian Fainelli most recently reported that xfsprogs doesn't build with musl on mips:
"MIPS platforms building with recent kernel headers and the musl-libc toolchain will expose the following build failure:
mmap.c: In function 'mmap_f':
mmap.c:196:12: error: 'MAP_SYNC' undeclared (first use in this function); did you mean 'MS_SYNC'?196 | flags = MAP_SYNC | MAP_SHARED_VALIDATE; | ^~~~~~~~ | MS_SYNCmmap.c:196:12: note: each undeclared identifier is reported only once for each function it appears in make[4]: *** [../include/buildrules:81: mmap.o] Error 1"
At first glance, the build failure here is caused by the fact that:
1. The configure script doesn't detect
MAP_SYNCsupport
2. The build system doesn't setHAVE_MAP_SYNC
2.io/mmap.cincludesinput.h->projects.h->xfs.hand latersys/mman.h
3.include/linux.h#define'sMAP_SYNCto 0 ifHAVE_MAP_SYNCis not set
4. musl'ssys/mman.h#undefMAP_SYNCon platforms that don't support it
5.io/mmap.ctries to useMAP_SYNC, not realizing that libc undefined itNormally,
xfs_ioonly exports functionality that is defined by the libc and/or kernel headers on the build system. We often make exceptions for new functionality so that we have a way to test them before the header file packages catch up, hence this '#ifndefHAVE_FOO#define FOO' paradigm.
MAP_SYNCis a gross and horribly broken example of this. These support crutches are supposed to be *private* to xfsprogs for benefit of early testing, but they were instead added toinclude/linux.h, which we provide to user programs in the xfslibs-dev package. IOWs, we've been #definingMAP_SYNCto zero for unsuspecting programs.Worst yet, gcc 11.3 doesn't even warn about overriding a #define to 0:
#include <
stdio.h>
#include <sys/mman.h>
#ifdef STUPID# include <xfs/xfs.h> #endifint main(int argc, char *argv[]) { printf("MAP_SYNC 0x%x\n", MAP_SYNC); }$ gcc -o a a.c -Wall $ ./a MAP_SYNC 0x80000 $ gcc -DSTUPID -o a a.c -Wall $ ./a MAP_SYNC 0x0Four years have gone by since the introduction of
MAP_SYNC, so let's get rid of the override code entirely -- any platform that supportsMAP_SYNChas had plenty of chances to ensure their header files have the right bits. While we're at it, fixAC_HAVE_MAP_SYNCto look forMAP_SYNCin the same header file that the one user (io/mmap.c) uses --sys/mman.h.Annoyingly, I had to test this by hand because the sole fstest that exercises
MAP_SYNC(generic/470) requires dm-logwrites and dm-thinp, neither of which support fsdax on current kernels.Reported-by [email protected]Reported-by Fabrice Fontaine <[email protected]>Reported-by Florian Fainelli <[email protected]>Signed-off-by Darrick J. Wong <[email protected]>--- include/linux.h | 8 -------- io/io.h | 2 +- io/mmap.c | 25 +++++++++++++------------ m4/package_libcdev.m4 | 3 +-- 4 files changed, 15 insertions(+), 23 deletions(-)include/linux.h +0
-8@@ -251,14 +251,6 @@ struct fsxattr {#define FS_XFLAG_COWEXTSIZE 0x00010000/* CoW extent size allocator hint */ #endif -#ifndef HAVE_MAP_SYNC -#define MAP_SYNC 0 -#define MAP_SHARED_VALIDATE 0 -#else -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> -#endif /* HAVE_MAP_SYNC */ -/* * Reminder: anything added to this file will be compiled into downstream * userspace projects! io/io.h +1
-1@@ -58,7 +58,7 @@ typedef struct mmap_region {size_t length;/* length of mapping */ off64_t offset;/* start offset into backing file */ int prot;/* protection mode of the mapping */ -bool map_sync;/* is this a MAP_SYNC mapping? */ +int flags;/* MAP_* flags passed to mmap() */ char *name;/* name of backing file */ } mmap_region_t;io/mmap.c +13
-12@@ -46,8 +46,11 @@ print_mapping(for (i =0 , p = pflags; p->prot !=PROT_NONE ; i++, p++) buffer[i] = (map->prot & p->prot) ? p->mode :'-' ; -if (map->map_sync) +#ifdef HAVE_MAP_SYNC +if ((map->flags & (MAP_SYNC |MAP_SHARED_VALIDATE )) == + (MAP_SYNC |MAP_SHARED_VALIDATE ))sprintf (&buffer[i]," S" ); +#endif printf("%c%03d%c 0x%lx - 0x%lx %s %14s (%lld : %ld)\n", braces? '[' : ' ', index, braces? ']' : ' ',@@ -139,7 +142,9 @@ mmap_help(void) " -r -- map with PROT_READ protection\n" " -w -- map with PROT_WRITE protection\n" " -x -- map with PROT_EXEC protection\n" +#ifdef HAVE_MAP_SYNC " -S -- map with MAP_SYNC and MAP_SHARED_VALIDATE flags\n" +#endif" -s <size> -- first do mmap(size)/munmap(size), try to reserve some free space\n" " If no protection mode is specified, all are used by default.\n" "\n"));@@ -193,18 +198,14 @@ mmap_f( prot |=PROT_EXEC ;break ;case 'S' : +#ifdef HAVE_MAP_SYNC flags =MAP_SYNC |MAP_SHARED_VALIDATE ; - -/* -* If MAP_SYNC and MAP_SHARED_VALIDATE aren't defined -* in the system headers we will have defined them -* both as 0. -*/ -if (!flags) { -printf ("MAP_SYNC not supported\n" ); -return 0 ; - }break ; +#else +printf ("MAP_SYNC not supported\n" ); + exitcode =1 ; +return command_usage (&mmap_cmd); +#endif case 's' : length2 =cvtnum (blocksize, sectsize, optarg);break ;@@ -281,7 +282,7 @@ mmap_f( mapping->offset = offset; mapping->name = filename; mapping->prot = prot; - mapping->map_sync = (flags == (MAP_SYNC |MAP_SHARED_VALIDATE )); + mapping->flags = flags;return 0 ; }m4/package_libcdev.m4 +0
-0m4/package_libcdev.m4 AC_DEFUN([AC_HAVE_MAP_SYNC], · line 393
Quoted 9 lines
m4/package_libcdev.m4 +1
-2@@ -387,8 +387,7 @@ AC_DEFUN([AC_HAVE_MAP_SYNC], [ AC_MSG_CHECKING([for MAP_SYNC]) AC_COMPILE_IFELSE( [ AC_LANG_PROGRAM([[ -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> +#include <sys/mman.h> ]], [[ int flags = MAP_SYNC | MAP_SHARED_VALIDATE; ]])Patch looks good, if you plan to make it into a non-RFC patch, feel free to include:
Quoted 2 lines
Reviewed-by Carlos Maiolino <[email protected]>
Darrick, do you need to re-post, or can the maintainers pick up the patch directly?
Signature · 1 line
Florian
Quoted 196 lines · Florian Fainelli
Quoted 195 lines · Carlos Maiolino
Quoted 181 lines · Darrick J. Wong
Can one of you please apply this patch and see if it'll build in musl on mips, please? Sorry it's taken so long to address this. :/
--D
---
From: Darrick J. Wong <[email protected]>Florian Fainelli most recently reported that xfsprogs doesn't build with musl on mips:
"MIPS platforms building with recent kernel headers and the musl-libc toolchain will expose the following build failure:
mmap.c: In function 'mmap_f':
mmap.c:196:12: error: 'MAP_SYNC' undeclared (first use in this function); did you mean 'MS_SYNC'?196 | flags = MAP_SYNC | MAP_SHARED_VALIDATE; | ^~~~~~~~ | MS_SYNCmmap.c:196:12: note: each undeclared identifier is reported only once for each function it appears in make[4]: *** [../include/buildrules:81: mmap.o] Error 1"
At first glance, the build failure here is caused by the fact that:
1. The configure script doesn't detect
MAP_SYNCsupport
2. The build system doesn't setHAVE_MAP_SYNC
2.io/mmap.cincludesinput.h->projects.h->xfs.hand latersys/mman.h
3.include/linux.h#define'sMAP_SYNCto 0 ifHAVE_MAP_SYNCis not set
4. musl'ssys/mman.h#undefMAP_SYNCon platforms that don't support it
5.io/mmap.ctries to useMAP_SYNC, not realizing that libc undefined itNormally,
xfs_ioonly exports functionality that is defined by the libc and/or kernel headers on the build system. We often make exceptions for new functionality so that we have a way to test them before the header file packages catch up, hence this '#ifndefHAVE_FOO#define FOO' paradigm.
MAP_SYNCis a gross and horribly broken example of this. These support crutches are supposed to be *private* to xfsprogs for benefit of early testing, but they were instead added toinclude/linux.h, which we provide to user programs in the xfslibs-dev package. IOWs, we've been #definingMAP_SYNCto zero for unsuspecting programs.Worst yet, gcc 11.3 doesn't even warn about overriding a #define to 0:
#include <
stdio.h>
#include <sys/mman.h>
#ifdef STUPID# include <xfs/xfs.h> #endifint main(int argc, char *argv[]) { printf("MAP_SYNC 0x%x\n", MAP_SYNC); }$ gcc -o a a.c -Wall $ ./a MAP_SYNC 0x80000 $ gcc -DSTUPID -o a a.c -Wall $ ./a MAP_SYNC 0x0Four years have gone by since the introduction of
MAP_SYNC, so let's get rid of the override code entirely -- any platform that supportsMAP_SYNChas had plenty of chances to ensure their header files have the right bits. While we're at it, fixAC_HAVE_MAP_SYNCto look forMAP_SYNCin the same header file that the one user (io/mmap.c) uses --sys/mman.h.Annoyingly, I had to test this by hand because the sole fstest that exercises
MAP_SYNC(generic/470) requires dm-logwrites and dm-thinp, neither of which support fsdax on current kernels.Reported-by [email protected]Reported-by Fabrice Fontaine <[email protected]>Reported-by Florian Fainelli <[email protected]>Signed-off-by Darrick J. Wong <[email protected]>--- include/linux.h | 8 -------- io/io.h | 2 +- io/mmap.c | 25 +++++++++++++------------ m4/package_libcdev.m4 | 3 +-- 4 files changed, 15 insertions(+), 23 deletions(-)include/linux.h +0
-8@@ -251,14 +251,6 @@ struct fsxattr {#define FS_XFLAG_COWEXTSIZE 0x00010000/* CoW extent size allocator hint */ #endif -#ifndef HAVE_MAP_SYNC -#define MAP_SYNC 0 -#define MAP_SHARED_VALIDATE 0 -#else -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> -#endif /* HAVE_MAP_SYNC */ -/* * Reminder: anything added to this file will be compiled into downstream * userspace projects! io/io.h +1
-1@@ -58,7 +58,7 @@ typedef struct mmap_region {size_t length;/* length of mapping */ off64_t offset;/* start offset into backing file */ int prot;/* protection mode of the mapping */ -bool map_sync;/* is this a MAP_SYNC mapping? */ +int flags;/* MAP_* flags passed to mmap() */ char *name;/* name of backing file */ } mmap_region_t;io/mmap.c +13
-12@@ -46,8 +46,11 @@ print_mapping(for (i =0 , p = pflags; p->prot !=PROT_NONE ; i++, p++) buffer[i] = (map->prot & p->prot) ? p->mode :'-' ; -if (map->map_sync) +#ifdef HAVE_MAP_SYNC +if ((map->flags & (MAP_SYNC |MAP_SHARED_VALIDATE )) == + (MAP_SYNC |MAP_SHARED_VALIDATE ))sprintf (&buffer[i]," S" ); +#endif printf("%c%03d%c 0x%lx - 0x%lx %s %14s (%lld : %ld)\n", braces? '[' : ' ', index, braces? ']' : ' ',@@ -139,7 +142,9 @@ mmap_help(void) " -r -- map with PROT_READ protection\n" " -w -- map with PROT_WRITE protection\n" " -x -- map with PROT_EXEC protection\n" +#ifdef HAVE_MAP_SYNC " -S -- map with MAP_SYNC and MAP_SHARED_VALIDATE flags\n" +#endif" -s <size> -- first do mmap(size)/munmap(size), try to reserve some free space\n" " If no protection mode is specified, all are used by default.\n" "\n"));@@ -193,18 +198,14 @@ mmap_f( prot |=PROT_EXEC ;break ;case 'S' : +#ifdef HAVE_MAP_SYNC flags =MAP_SYNC |MAP_SHARED_VALIDATE ; - -/* -* If MAP_SYNC and MAP_SHARED_VALIDATE aren't defined -* in the system headers we will have defined them -* both as 0. -*/ -if (!flags) { -printf ("MAP_SYNC not supported\n" ); -return 0 ; - }break ; +#else +printf ("MAP_SYNC not supported\n" ); + exitcode =1 ; +return command_usage (&mmap_cmd); +#endif case 's' : length2 =cvtnum (blocksize, sectsize, optarg);break ;@@ -281,7 +282,7 @@ mmap_f( mapping->offset = offset; mapping->name = filename; mapping->prot = prot; - mapping->map_sync = (flags == (MAP_SYNC |MAP_SHARED_VALIDATE )); + mapping->flags = flags;return 0 ; }m4/package_libcdev.m4 +0
-0m4/package_libcdev.m4 AC_DEFUN([AC_HAVE_MAP_SYNC], · line 393
Quoted 9 lines
m4/package_libcdev.m4 +1
-2@@ -387,8 +387,7 @@ AC_DEFUN([AC_HAVE_MAP_SYNC], [ AC_MSG_CHECKING([for MAP_SYNC]) AC_COMPILE_IFELSE( [ AC_LANG_PROGRAM([[ -#include <asm-generic/mman.h> -#include <asm-generic/mman-common.h> +#include <sys/mman.h> ]], [[ int flags = MAP_SYNC | MAP_SHARED_VALIDATE; ]])Patch looks good, if you plan to make it into a non-RFC patch, feel free to include:
Quoted 4 lines
Quoted 2 lines
Reviewed-by Carlos Maiolino <[email protected]>Darrick, do you need to re-post, or can the maintainers pick up the patch directly?
I already did:
https://lore.kernel.org/linux-xfs/YtmB005kkkErb5uw@magnolia/
(It's August, so I think the xfsprogs maintainer might be on vacation? Either way, I'll make sure he's aware of it before the next release.)
--D
Quoted 2 lines
Signature · 1 line
Florian
Quoted 9 lines · Darrick J. Wong
Quoted 1 line
Darrick, do you need to re-post, or can the maintainers pick up the patch directly?
I already did:
https://lore.kernel.org/linux-xfs/YtmB005kkkErb5uw@magnolia/(It's August, so I think the xfsprogs maintainer might be on vacation? Either way, I'll make sure he's aware of it before the next release.)
--D
Yep I was, picking it up now. Thanks for the pointer Darrick.
-Eric
Quoted 10 lines · Eric Sandeen
Quoted 7 lines · Darrick J. Wong
Quoted 1 line
Darrick, do you need to re-post, or can the maintainers pick up the patch directly?
I already did:
https://lore.kernel.org/linux-xfs/YtmB005kkkErb5uw@magnolia/(It's August, so I think the xfsprogs maintainer might be on vacation? Either way, I'll make sure he's aware of it before the next release.)
Quoted 6 lines
Quoted 2 lines
--D
Yep I was, picking it up now. Thanks for the pointer Darrick.
-Eric
Eric, any chance this could be pushed to xfsprogs-dev.git so we can take the patch and submit it to buildroot, OpenWrt and other projects that depend upon that build fix?
Thanks!
Signature · 1 line
Florian
Quoted 14 lines · Florian Fainelli
Quoted 13 lines · Eric Sandeen
Quoted 9 lines · Darrick J. Wong
Quoted 1 line
Darrick, do you need to re-post, or can the maintainers pick up the patch directly?
I already did:
https://lore.kernel.org/linux-xfs/YtmB005kkkErb5uw@magnolia/(It's August, so I think the xfsprogs maintainer might be on vacation? Either way, I'll make sure he's aware of it before the next release.)
--D
Quoted 7 lines
Quoted 3 lines
Yep I was, picking it up now. Thanks for the pointer Darrick.
-Eric
Eric, any chance this could be pushed to xfsprogs-dev.git so we can take the patch and submit it to buildroot, OpenWrt and other projects that depend upon that build fix?
It's queued in for-next, so the git commit ids should be stable if you want to get that started now.
--D
Quoted 3 lines
Thanks!
Signature · 1 line
Florian
Quoted 19 lines · Darrick J. Wong
Quoted 18 lines · Florian Fainelli
Quoted 16 lines · Eric Sandeen
Quoted 9 lines · Darrick J. Wong
Quoted 1 line
Darrick, do you need to re-post, or can the maintainers pick up the patch directly?
I already did:
https://lore.kernel.org/linux-xfs/YtmB005kkkErb5uw@magnolia/(It's August, so I think the xfsprogs maintainer might be on vacation? Either way, I'll make sure he's aware of it before the next release.)
--D
Yep I was, picking it up now. Thanks for the pointer Darrick.
-Eric
Quoted 8 lines
Quoted 3 lines
Eric, any chance this could be pushed to xfsprogs-dev.git so we can take the patch and submit it to buildroot, OpenWrt and other projects that depend upon that build fix?
It's queued in for-next, so the git commit ids should be stable if you want to get that started now.
Woah, sorry for not noticing, thanks!
Signature · 1 line
Florian