| Message ID | 20231024201852.1512720-1-chantr4@gmail.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | BPF |
| Headers | show |
| Series | [bpf-next] selftests/bpf: umount children of TDIR in test_bpffs | expand |
On 10/24/23 13:18, Manu Bretelle wrote: > Currently this tests tries to umount /sys/kernel/debug (TDIR) but the > system it is running on may have mounts below. > > For example, danobi/vmtest [0] VMs have > mount -t tracefs tracefs /sys/kernel/debug/tracing > as part of their init. > > This change list mounts and will umount any mounts below TDIR before > umounting TDIR itself. > > Note that it is not umounting recursively, so in the case of a sub-mount > of TDIR having another sub-mount, this will fail as mtab is ordered. Should we move TID to a random path likes "/sys/kernel/debug-<pid>/"? > > Test: > > Originally: > > $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" > => bzImage > ===> Booting > ===> Setting up VM > ===> Running command > [ 2.138818] bpf_testmod: loading out-of-tree module taints kernel. > [ 2.140913] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel > bpf_testmod.ko is already unloaded. > Loading bpf_testmod.ko... > Successfully loaded bpf_testmod.ko. > test_test_bpffs:PASS:clone 0 nsec > fn:PASS:unshare 0 nsec > fn:PASS:mount / 0 nsec > fn:FAIL:umount /sys/kernel/debug unexpected error: -1 (errno 16) > bpf_testmod.ko is already unloaded. > Loading bpf_testmod.ko... > Successfully loaded bpf_testmod.ko. > test_test_bpffs:PASS:clone 0 nsec > test_test_bpffs:PASS:waitpid 0 nsec > test_test_bpffs:FAIL:bpffs test failed 255#282 test_bpffs:FAIL > Summary: 0/0 PASSED, 0 SKIPPED, 1 FAILED > Successfully unloaded bpf_testmod.ko. > Command failed with exit code: 1 > > After this change: > > $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" > => bzImage > ===> Booting > ===> Setting up VM > ===> Running command > [ 2.035210] bpf_testmod: loading out-of-tree module taints kernel. > [ 2.036510] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel > bpf_testmod.ko is already unloaded. > Loading bpf_testmod.ko... > Successfully loaded bpf_testmod.ko. > test_test_bpffs:PASS:clone 0 nsec > fn:PASS:unshare 0 nsec > fn:PASS:mount / 0 nsec > fn:PASS:accessing /etc/mtab 0 nsec > fn:PASS:umount /sys/kernel/debug/tracing 0 nsec > fn:PASS:umount /sys/kernel/debug 0 nsec > fn:PASS:mount tmpfs 0 nsec > fn:PASS:mkdir /sys/kernel/debug/fs1 0 nsec > fn:PASS:mkdir /sys/kernel/debug/fs2 0 nsec > fn:PASS:mount bpffs /sys/kernel/debug/fs1 0 nsec > fn:PASS:mount bpffs /sys/kernel/debug/fs2 0 nsec > fn:PASS:reading /sys/kernel/debug/fs1/maps.debug 0 nsec > fn:PASS:reading /sys/kernel/debug/fs2/progs.debug 0 nsec > fn:PASS:creating /sys/kernel/debug/fs1/a 0 nsec > fn:PASS:creating /sys/kernel/debug/fs1/a/1 0 nsec > fn:PASS:creating /sys/kernel/debug/fs1/b 0 nsec > fn:PASS:create_map(ARRAY) 0 nsec > fn:PASS:pin map 0 nsec > fn:PASS:stat(/sys/kernel/debug/fs1/a) 0 nsec > fn:PASS:renameat2(/fs1/a, /fs1/b, RENAME_EXCHANGE) 0 nsec > fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec > fn:PASS:b should have a's inode 0 nsec > fn:PASS:access(/sys/kernel/debug/fs1/b/1) 0 nsec > fn:PASS:stat(/sys/kernel/debug/fs1/map) 0 nsec > fn:PASS:renameat2(/fs1/c, /fs1/b, RENAME_EXCHANGE) 0 nsec > fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec > fn:PASS:b should have c's inode 0 nsec > fn:PASS:access(/sys/kernel/debug/fs1/c/1) 0 nsec > fn:PASS:renameat2(RENAME_NOREPLACE) 0 nsec > fn:PASS:access(/sys/kernel/debug/fs1/b) 0 nsec > bpf_testmod.ko is already unloaded. > Loading bpf_testmod.ko... > Successfully loaded bpf_testmod.ko. > test_test_bpffs:PASS:clone 0 nsec > test_test_bpffs:PASS:waitpid 0 nsec > test_test_bpffs:PASS:bpffs test 0 nsec > #282 test_bpffs:OK > Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED > Successfully unloaded bpf_testmod.ko. > > [0] https://github.com/danobi/vmtest > > Signed-off-by: Manu Bretelle <chantr4@gmail.com> > --- > .../selftests/bpf/prog_tests/test_bpffs.c | 28 +++++++++++++++++++ > 1 file changed, 28 insertions(+) > > diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > index 214d9f4a94a5..001bf694c269 100644 > --- a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > +++ b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > @@ -3,12 +3,14 @@ > #define _GNU_SOURCE > #include <stdio.h> > #include <sched.h> > +#include <mntent.h> > #include <sys/mount.h> > #include <sys/stat.h> > #include <sys/types.h> > #include <test_progs.h> > > #define TDIR "/sys/kernel/debug" > +#define MTAB "/etc/mtab" > > static int read_iter(char *file) > { > @@ -32,6 +34,8 @@ static int read_iter(char *file) > > static int fn(void) > { > + /* A buffer to store logging messages */ > + char buf[1024]; > struct stat a, b, c; > int err, map; > > @@ -43,6 +47,30 @@ static int fn(void) > if (!ASSERT_OK(err, "mount /")) > goto out; > > + /* TDIR may have mounts below. unount them first */ > + FILE *mtab = setmntent(MTAB, "r"); > + > + if (!ASSERT_TRUE(mtab != NULL, "accessing " MTAB)) { > + err = errno; > + goto out; > + } > + > + struct mntent *mnt = NULL; > + > + while ((mnt = getmntent(mtab)) != NULL) { > + if (strlen(mnt->mnt_dir) > strlen(TDIR) && > + strncmp(TDIR, mnt->mnt_dir, strlen(TDIR)) == 0) { > + snprintf(buf, sizeof(buf) - 1, "umount %s", mnt->mnt_dir); > + err = umount(mnt->mnt_dir); > + if (!ASSERT_OK(err, buf)) { > + endmntent(mtab); > + goto out; > + } > + } > + } > + // Ignore any error here > + endmntent(mtab); > + > err = umount(TDIR); > if (!ASSERT_OK(err, "umount " TDIR)) > goto out;
On Tue, Oct 24, 2023 at 02:29:19PM -0700, Kui-Feng Lee wrote: > > > On 10/24/23 13:18, Manu Bretelle wrote: > > Currently this tests tries to umount /sys/kernel/debug (TDIR) but the > > system it is running on may have mounts below. > > > > For example, danobi/vmtest [0] VMs have > > mount -t tracefs tracefs /sys/kernel/debug/tracing > > as part of their init. > > > > This change list mounts and will umount any mounts below TDIR before > > umounting TDIR itself. > > > > Note that it is not umounting recursively, so in the case of a sub-mount > > of TDIR having another sub-mount, this will fail as mtab is ordered. > > Should we move TID to a random path likes "/sys/kernel/debug-<pid>/"? > Fair point, I suppose we would want to keep TDIR a defined string as it does simplify the gymnastic involved through the rest of the script, but yeah looking at the original commit: edb65ee5aa25 (selftests/bpf: Add bpffs preload test) I don't see any reason to use an alternate directory and rather mkdir it vs umounting the original one. so something like #define TDIR "/sys/kernel/test_bpffs" Would probably do. Alexei could confirm his original intent probably. > > > > Test: > > > > Originally: > > > > $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" > > => bzImage > > ===> Booting > > ===> Setting up VM > > ===> Running command > > [ 2.138818] bpf_testmod: loading out-of-tree module taints kernel. > > [ 2.140913] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel > > bpf_testmod.ko is already unloaded. > > Loading bpf_testmod.ko... > > Successfully loaded bpf_testmod.ko. > > test_test_bpffs:PASS:clone 0 nsec > > fn:PASS:unshare 0 nsec > > fn:PASS:mount / 0 nsec > > fn:FAIL:umount /sys/kernel/debug unexpected error: -1 (errno 16) > > bpf_testmod.ko is already unloaded. > > Loading bpf_testmod.ko... > > Successfully loaded bpf_testmod.ko. > > test_test_bpffs:PASS:clone 0 nsec > > test_test_bpffs:PASS:waitpid 0 nsec > > test_test_bpffs:FAIL:bpffs test failed 255#282 test_bpffs:FAIL > > Summary: 0/0 PASSED, 0 SKIPPED, 1 FAILED > > Successfully unloaded bpf_testmod.ko. > > Command failed with exit code: 1 > > > > After this change: > > > > $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" > > => bzImage > > ===> Booting > > ===> Setting up VM > > ===> Running command > > [ 2.035210] bpf_testmod: loading out-of-tree module taints kernel. > > [ 2.036510] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel > > bpf_testmod.ko is already unloaded. > > Loading bpf_testmod.ko... > > Successfully loaded bpf_testmod.ko. > > test_test_bpffs:PASS:clone 0 nsec > > fn:PASS:unshare 0 nsec > > fn:PASS:mount / 0 nsec > > fn:PASS:accessing /etc/mtab 0 nsec > > fn:PASS:umount /sys/kernel/debug/tracing 0 nsec > > fn:PASS:umount /sys/kernel/debug 0 nsec > > fn:PASS:mount tmpfs 0 nsec > > fn:PASS:mkdir /sys/kernel/debug/fs1 0 nsec > > fn:PASS:mkdir /sys/kernel/debug/fs2 0 nsec > > fn:PASS:mount bpffs /sys/kernel/debug/fs1 0 nsec > > fn:PASS:mount bpffs /sys/kernel/debug/fs2 0 nsec > > fn:PASS:reading /sys/kernel/debug/fs1/maps.debug 0 nsec > > fn:PASS:reading /sys/kernel/debug/fs2/progs.debug 0 nsec > > fn:PASS:creating /sys/kernel/debug/fs1/a 0 nsec > > fn:PASS:creating /sys/kernel/debug/fs1/a/1 0 nsec > > fn:PASS:creating /sys/kernel/debug/fs1/b 0 nsec > > fn:PASS:create_map(ARRAY) 0 nsec > > fn:PASS:pin map 0 nsec > > fn:PASS:stat(/sys/kernel/debug/fs1/a) 0 nsec > > fn:PASS:renameat2(/fs1/a, /fs1/b, RENAME_EXCHANGE) 0 nsec > > fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec > > fn:PASS:b should have a's inode 0 nsec > > fn:PASS:access(/sys/kernel/debug/fs1/b/1) 0 nsec > > fn:PASS:stat(/sys/kernel/debug/fs1/map) 0 nsec > > fn:PASS:renameat2(/fs1/c, /fs1/b, RENAME_EXCHANGE) 0 nsec > > fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec > > fn:PASS:b should have c's inode 0 nsec > > fn:PASS:access(/sys/kernel/debug/fs1/c/1) 0 nsec > > fn:PASS:renameat2(RENAME_NOREPLACE) 0 nsec > > fn:PASS:access(/sys/kernel/debug/fs1/b) 0 nsec > > bpf_testmod.ko is already unloaded. > > Loading bpf_testmod.ko... > > Successfully loaded bpf_testmod.ko. > > test_test_bpffs:PASS:clone 0 nsec > > test_test_bpffs:PASS:waitpid 0 nsec > > test_test_bpffs:PASS:bpffs test 0 nsec > > #282 test_bpffs:OK > > Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED > > Successfully unloaded bpf_testmod.ko. > > > > [0] https://github.com/danobi/vmtest > > > > Signed-off-by: Manu Bretelle <chantr4@gmail.com> > > --- > > .../selftests/bpf/prog_tests/test_bpffs.c | 28 +++++++++++++++++++ > > 1 file changed, 28 insertions(+) > > > > diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > > index 214d9f4a94a5..001bf694c269 100644 > > --- a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > > +++ b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > > @@ -3,12 +3,14 @@ > > #define _GNU_SOURCE > > #include <stdio.h> > > #include <sched.h> > > +#include <mntent.h> > > #include <sys/mount.h> > > #include <sys/stat.h> > > #include <sys/types.h> > > #include <test_progs.h> > > #define TDIR "/sys/kernel/debug" > > +#define MTAB "/etc/mtab" > > static int read_iter(char *file) > > { > > @@ -32,6 +34,8 @@ static int read_iter(char *file) > > static int fn(void) > > { > > + /* A buffer to store logging messages */ > > + char buf[1024]; > > struct stat a, b, c; > > int err, map; > > @@ -43,6 +47,30 @@ static int fn(void) > > if (!ASSERT_OK(err, "mount /")) > > goto out; > > + /* TDIR may have mounts below. unount them first */ > > + FILE *mtab = setmntent(MTAB, "r"); > > + > > + if (!ASSERT_TRUE(mtab != NULL, "accessing " MTAB)) { > > + err = errno; > > + goto out; > > + } > > + > > + struct mntent *mnt = NULL; > > + > > + while ((mnt = getmntent(mtab)) != NULL) { > > + if (strlen(mnt->mnt_dir) > strlen(TDIR) && > > + strncmp(TDIR, mnt->mnt_dir, strlen(TDIR)) == 0) { > > + snprintf(buf, sizeof(buf) - 1, "umount %s", mnt->mnt_dir); > > + err = umount(mnt->mnt_dir); > > + if (!ASSERT_OK(err, buf)) { > > + endmntent(mtab); > > + goto out; > > + } > > + } > > + } > > + // Ignore any error here > > + endmntent(mtab); > > + > > err = umount(TDIR); > > if (!ASSERT_OK(err, "umount " TDIR)) > > goto out;
On Tue, Oct 24, 2023 at 10:42 PM Manu Bretelle <chantr4@gmail.com> wrote: > > On Tue, Oct 24, 2023 at 02:29:19PM -0700, Kui-Feng Lee wrote: > > > > > > On 10/24/23 13:18, Manu Bretelle wrote: > > > Currently this tests tries to umount /sys/kernel/debug (TDIR) but the > > > system it is running on may have mounts below. > > > > > > For example, danobi/vmtest [0] VMs have > > > mount -t tracefs tracefs /sys/kernel/debug/tracing > > > as part of their init. > > > > > > This change list mounts and will umount any mounts below TDIR before > > > umounting TDIR itself. > > > > > > Note that it is not umounting recursively, so in the case of a sub-mount > > > of TDIR having another sub-mount, this will fail as mtab is ordered. > > > > Should we move TID to a random path likes "/sys/kernel/debug-<pid>/"? > > > > Fair point, I suppose we would want to keep TDIR a defined string as it does > simplify the gymnastic involved through the rest of the script, but yeah > looking at the original commit: > edb65ee5aa25 (selftests/bpf: Add bpffs preload test) > > I don't see any reason to use an alternate directory and rather mkdir it vs > umounting the original one. > so something like > > #define TDIR "/sys/kernel/test_bpffs" > > Would probably do. > > Alexei could confirm his original intent probably. I don't remember why I picked /sys/kernel/debug back then. I suspect TDIR /tmp/foo and mkdir would work the same way.
On 10/24/23 10:41 PM, Manu Bretelle wrote: > On Tue, Oct 24, 2023 at 02:29:19PM -0700, Kui-Feng Lee wrote: >> >> On 10/24/23 13:18, Manu Bretelle wrote: >>> Currently this tests tries to umount /sys/kernel/debug (TDIR) but the >>> system it is running on may have mounts below. >>> >>> For example, danobi/vmtest [0] VMs have >>> mount -t tracefs tracefs /sys/kernel/debug/tracing >>> as part of their init. >>> >>> This change list mounts and will umount any mounts below TDIR before >>> umounting TDIR itself. >>> >>> Note that it is not umounting recursively, so in the case of a sub-mount >>> of TDIR having another sub-mount, this will fail as mtab is ordered. >> Should we move TID to a random path likes "/sys/kernel/debug-<pid>/"? >> > Fair point, I suppose we would want to keep TDIR a defined string as it does > simplify the gymnastic involved through the rest of the script, but yeah > looking at the original commit: > edb65ee5aa25 (selftests/bpf: Add bpffs preload test) > > I don't see any reason to use an alternate directory and rather mkdir it vs > umounting the original one. > so something like > > #define TDIR "/sys/kernel/test_bpffs" > > Would probably do. > > Alexei could confirm his original intent probably. Maybe/sys/kernel/tracing should work too? Not sure whether it is universally available or not. > >>> Test: >>> >>> Originally: >>> >>> $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" >>> => bzImage >>> ===> Booting >>> ===> Setting up VM >>> ===> Running command >>> [ 2.138818] bpf_testmod: loading out-of-tree module taints kernel. >>> [ 2.140913] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel >>> bpf_testmod.ko is already unloaded. >>> Loading bpf_testmod.ko... >>> Successfully loaded bpf_testmod.ko. >>> test_test_bpffs:PASS:clone 0 nsec >>> fn:PASS:unshare 0 nsec >>> fn:PASS:mount / 0 nsec >>> fn:FAIL:umount /sys/kernel/debug unexpected error: -1 (errno 16) >>> bpf_testmod.ko is already unloaded. >>> Loading bpf_testmod.ko... >>> Successfully loaded bpf_testmod.ko. >>> test_test_bpffs:PASS:clone 0 nsec >>> test_test_bpffs:PASS:waitpid 0 nsec >>> test_test_bpffs:FAIL:bpffs test failed 255#282 test_bpffs:FAIL >>> Summary: 0/0 PASSED, 0 SKIPPED, 1 FAILED >>> Successfully unloaded bpf_testmod.ko. >>> Command failed with exit code: 1 >>> >>> After this change: >>> >>> $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" >>> => bzImage >>> ===> Booting >>> ===> Setting up VM >>> ===> Running command >>> [ 2.035210] bpf_testmod: loading out-of-tree module taints kernel. >>> [ 2.036510] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel >>> bpf_testmod.ko is already unloaded. >>> Loading bpf_testmod.ko... >>> Successfully loaded bpf_testmod.ko. >>> test_test_bpffs:PASS:clone 0 nsec >>> fn:PASS:unshare 0 nsec >>> fn:PASS:mount / 0 nsec >>> fn:PASS:accessing /etc/mtab 0 nsec >>> fn:PASS:umount /sys/kernel/debug/tracing 0 nsec >>> fn:PASS:umount /sys/kernel/debug 0 nsec >>> fn:PASS:mount tmpfs 0 nsec >>> fn:PASS:mkdir /sys/kernel/debug/fs1 0 nsec >>> fn:PASS:mkdir /sys/kernel/debug/fs2 0 nsec >>> fn:PASS:mount bpffs /sys/kernel/debug/fs1 0 nsec >>> fn:PASS:mount bpffs /sys/kernel/debug/fs2 0 nsec >>> fn:PASS:reading /sys/kernel/debug/fs1/maps.debug 0 nsec >>> fn:PASS:reading /sys/kernel/debug/fs2/progs.debug 0 nsec >>> fn:PASS:creating /sys/kernel/debug/fs1/a 0 nsec >>> fn:PASS:creating /sys/kernel/debug/fs1/a/1 0 nsec >>> fn:PASS:creating /sys/kernel/debug/fs1/b 0 nsec >>> fn:PASS:create_map(ARRAY) 0 nsec >>> fn:PASS:pin map 0 nsec >>> fn:PASS:stat(/sys/kernel/debug/fs1/a) 0 nsec >>> fn:PASS:renameat2(/fs1/a, /fs1/b, RENAME_EXCHANGE) 0 nsec >>> fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec >>> fn:PASS:b should have a's inode 0 nsec >>> fn:PASS:access(/sys/kernel/debug/fs1/b/1) 0 nsec >>> fn:PASS:stat(/sys/kernel/debug/fs1/map) 0 nsec >>> fn:PASS:renameat2(/fs1/c, /fs1/b, RENAME_EXCHANGE) 0 nsec >>> fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec >>> fn:PASS:b should have c's inode 0 nsec >>> fn:PASS:access(/sys/kernel/debug/fs1/c/1) 0 nsec >>> fn:PASS:renameat2(RENAME_NOREPLACE) 0 nsec >>> fn:PASS:access(/sys/kernel/debug/fs1/b) 0 nsec >>> bpf_testmod.ko is already unloaded. >>> Loading bpf_testmod.ko... >>> Successfully loaded bpf_testmod.ko. >>> test_test_bpffs:PASS:clone 0 nsec >>> test_test_bpffs:PASS:waitpid 0 nsec >>> test_test_bpffs:PASS:bpffs test 0 nsec >>> #282 test_bpffs:OK >>> Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED >>> Successfully unloaded bpf_testmod.ko. >>> >>> [0] https://github.com/danobi/vmtest >>> >>> Signed-off-by: Manu Bretelle <chantr4@gmail.com> >>> --- >>> .../selftests/bpf/prog_tests/test_bpffs.c | 28 +++++++++++++++++++ >>> 1 file changed, 28 insertions(+) >>> >>> diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c >>> index 214d9f4a94a5..001bf694c269 100644 >>> --- a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c >>> +++ b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c >>> @@ -3,12 +3,14 @@ >>> #define _GNU_SOURCE >>> #include <stdio.h> >>> #include <sched.h> >>> +#include <mntent.h> >>> #include <sys/mount.h> >>> #include <sys/stat.h> >>> #include <sys/types.h> >>> #include <test_progs.h> >>> #define TDIR "/sys/kernel/debug" >>> +#define MTAB "/etc/mtab" >>> static int read_iter(char *file) >>> { >>> @@ -32,6 +34,8 @@ static int read_iter(char *file) >>> static int fn(void) >>> { >>> + /* A buffer to store logging messages */ >>> + char buf[1024]; >>> struct stat a, b, c; >>> int err, map; >>> @@ -43,6 +47,30 @@ static int fn(void) >>> if (!ASSERT_OK(err, "mount /")) >>> goto out; >>> + /* TDIR may have mounts below. unount them first */ >>> + FILE *mtab = setmntent(MTAB, "r"); >>> + >>> + if (!ASSERT_TRUE(mtab != NULL, "accessing " MTAB)) { >>> + err = errno; >>> + goto out; >>> + } >>> + >>> + struct mntent *mnt = NULL; >>> + >>> + while ((mnt = getmntent(mtab)) != NULL) { >>> + if (strlen(mnt->mnt_dir) > strlen(TDIR) && >>> + strncmp(TDIR, mnt->mnt_dir, strlen(TDIR)) == 0) { >>> + snprintf(buf, sizeof(buf) - 1, "umount %s", mnt->mnt_dir); >>> + err = umount(mnt->mnt_dir); >>> + if (!ASSERT_OK(err, buf)) { >>> + endmntent(mtab); >>> + goto out; >>> + } >>> + } >>> + } >>> + // Ignore any error here >>> + endmntent(mtab); >>> + >>> err = umount(TDIR); >>> if (!ASSERT_OK(err, "umount " TDIR)) >>> goto out;
On Thu, Oct 26, 2023 at 11:08:35PM -0700, Yonghong Song wrote: > > On 10/24/23 10:41 PM, Manu Bretelle wrote: > > On Tue, Oct 24, 2023 at 02:29:19PM -0700, Kui-Feng Lee wrote: > > > > > > On 10/24/23 13:18, Manu Bretelle wrote: > > > > Currently this tests tries to umount /sys/kernel/debug (TDIR) but the > > > > system it is running on may have mounts below. > > > > > > > > For example, danobi/vmtest [0] VMs have > > > > mount -t tracefs tracefs /sys/kernel/debug/tracing > > > > as part of their init. > > > > > > > > This change list mounts and will umount any mounts below TDIR before > > > > umounting TDIR itself. > > > > > > > > Note that it is not umounting recursively, so in the case of a sub-mount > > > > of TDIR having another sub-mount, this will fail as mtab is ordered. > > > Should we move TID to a random path likes "/sys/kernel/debug-<pid>/"? > > > > > Fair point, I suppose we would want to keep TDIR a defined string as it does > > simplify the gymnastic involved through the rest of the script, but yeah > > looking at the original commit: > > edb65ee5aa25 (selftests/bpf: Add bpffs preload test) > > > > I don't see any reason to use an alternate directory and rather mkdir it vs > > umounting the original one. > > so something like > > > > #define TDIR "/sys/kernel/test_bpffs" > > > > Would probably do. > > > > Alexei could confirm his original intent probably. > > > Maybe/sys/kernel/tracing should work too? Not sure whether it is universally > available or not. > > I think we just need an existing directory (whichever) to mount tmpfs on, and Alexei previous comment seems to confirm that. But importantly, nothing that has sub-mounts if we don't want to get into the business of recursively umounting sub-mounts. I will just pick up a name which is random enough to not realistically exist on a system, mkdir it instead of the existing umount(TDIR) and we should be good. > > > > > > Test: > > > > > > > > Originally: > > > > > > > > $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" > > > > => bzImage > > > > ===> Booting > > > > ===> Setting up VM > > > > ===> Running command > > > > [ 2.138818] bpf_testmod: loading out-of-tree module taints kernel. > > > > [ 2.140913] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel > > > > bpf_testmod.ko is already unloaded. > > > > Loading bpf_testmod.ko... > > > > Successfully loaded bpf_testmod.ko. > > > > test_test_bpffs:PASS:clone 0 nsec > > > > fn:PASS:unshare 0 nsec > > > > fn:PASS:mount / 0 nsec > > > > fn:FAIL:umount /sys/kernel/debug unexpected error: -1 (errno 16) > > > > bpf_testmod.ko is already unloaded. > > > > Loading bpf_testmod.ko... > > > > Successfully loaded bpf_testmod.ko. > > > > test_test_bpffs:PASS:clone 0 nsec > > > > test_test_bpffs:PASS:waitpid 0 nsec > > > > test_test_bpffs:FAIL:bpffs test failed 255#282 test_bpffs:FAIL > > > > Summary: 0/0 PASSED, 0 SKIPPED, 1 FAILED > > > > Successfully unloaded bpf_testmod.ko. > > > > Command failed with exit code: 1 > > > > > > > > After this change: > > > > > > > > $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" > > > > => bzImage > > > > ===> Booting > > > > ===> Setting up VM > > > > ===> Running command > > > > [ 2.035210] bpf_testmod: loading out-of-tree module taints kernel. > > > > [ 2.036510] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel > > > > bpf_testmod.ko is already unloaded. > > > > Loading bpf_testmod.ko... > > > > Successfully loaded bpf_testmod.ko. > > > > test_test_bpffs:PASS:clone 0 nsec > > > > fn:PASS:unshare 0 nsec > > > > fn:PASS:mount / 0 nsec > > > > fn:PASS:accessing /etc/mtab 0 nsec > > > > fn:PASS:umount /sys/kernel/debug/tracing 0 nsec > > > > fn:PASS:umount /sys/kernel/debug 0 nsec > > > > fn:PASS:mount tmpfs 0 nsec > > > > fn:PASS:mkdir /sys/kernel/debug/fs1 0 nsec > > > > fn:PASS:mkdir /sys/kernel/debug/fs2 0 nsec > > > > fn:PASS:mount bpffs /sys/kernel/debug/fs1 0 nsec > > > > fn:PASS:mount bpffs /sys/kernel/debug/fs2 0 nsec > > > > fn:PASS:reading /sys/kernel/debug/fs1/maps.debug 0 nsec > > > > fn:PASS:reading /sys/kernel/debug/fs2/progs.debug 0 nsec > > > > fn:PASS:creating /sys/kernel/debug/fs1/a 0 nsec > > > > fn:PASS:creating /sys/kernel/debug/fs1/a/1 0 nsec > > > > fn:PASS:creating /sys/kernel/debug/fs1/b 0 nsec > > > > fn:PASS:create_map(ARRAY) 0 nsec > > > > fn:PASS:pin map 0 nsec > > > > fn:PASS:stat(/sys/kernel/debug/fs1/a) 0 nsec > > > > fn:PASS:renameat2(/fs1/a, /fs1/b, RENAME_EXCHANGE) 0 nsec > > > > fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec > > > > fn:PASS:b should have a's inode 0 nsec > > > > fn:PASS:access(/sys/kernel/debug/fs1/b/1) 0 nsec > > > > fn:PASS:stat(/sys/kernel/debug/fs1/map) 0 nsec > > > > fn:PASS:renameat2(/fs1/c, /fs1/b, RENAME_EXCHANGE) 0 nsec > > > > fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec > > > > fn:PASS:b should have c's inode 0 nsec > > > > fn:PASS:access(/sys/kernel/debug/fs1/c/1) 0 nsec > > > > fn:PASS:renameat2(RENAME_NOREPLACE) 0 nsec > > > > fn:PASS:access(/sys/kernel/debug/fs1/b) 0 nsec > > > > bpf_testmod.ko is already unloaded. > > > > Loading bpf_testmod.ko... > > > > Successfully loaded bpf_testmod.ko. > > > > test_test_bpffs:PASS:clone 0 nsec > > > > test_test_bpffs:PASS:waitpid 0 nsec > > > > test_test_bpffs:PASS:bpffs test 0 nsec > > > > #282 test_bpffs:OK > > > > Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED > > > > Successfully unloaded bpf_testmod.ko. > > > > > > > > [0] https://github.com/danobi/vmtest > > > > > > > > Signed-off-by: Manu Bretelle <chantr4@gmail.com> > > > > --- > > > > .../selftests/bpf/prog_tests/test_bpffs.c | 28 +++++++++++++++++++ > > > > 1 file changed, 28 insertions(+) > > > > > > > > diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > > > > index 214d9f4a94a5..001bf694c269 100644 > > > > --- a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > > > > +++ b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c > > > > @@ -3,12 +3,14 @@ > > > > #define _GNU_SOURCE > > > > #include <stdio.h> > > > > #include <sched.h> > > > > +#include <mntent.h> > > > > #include <sys/mount.h> > > > > #include <sys/stat.h> > > > > #include <sys/types.h> > > > > #include <test_progs.h> > > > > #define TDIR "/sys/kernel/debug" > > > > +#define MTAB "/etc/mtab" > > > > static int read_iter(char *file) > > > > { > > > > @@ -32,6 +34,8 @@ static int read_iter(char *file) > > > > static int fn(void) > > > > { > > > > + /* A buffer to store logging messages */ > > > > + char buf[1024]; > > > > struct stat a, b, c; > > > > int err, map; > > > > @@ -43,6 +47,30 @@ static int fn(void) > > > > if (!ASSERT_OK(err, "mount /")) > > > > goto out; > > > > + /* TDIR may have mounts below. unount them first */ > > > > + FILE *mtab = setmntent(MTAB, "r"); > > > > + > > > > + if (!ASSERT_TRUE(mtab != NULL, "accessing " MTAB)) { > > > > + err = errno; > > > > + goto out; > > > > + } > > > > + > > > > + struct mntent *mnt = NULL; > > > > + > > > > + while ((mnt = getmntent(mtab)) != NULL) { > > > > + if (strlen(mnt->mnt_dir) > strlen(TDIR) && > > > > + strncmp(TDIR, mnt->mnt_dir, strlen(TDIR)) == 0) { > > > > + snprintf(buf, sizeof(buf) - 1, "umount %s", mnt->mnt_dir); > > > > + err = umount(mnt->mnt_dir); > > > > + if (!ASSERT_OK(err, buf)) { > > > > + endmntent(mtab); > > > > + goto out; > > > > + } > > > > + } > > > > + } > > > > + // Ignore any error here > > > > + endmntent(mtab); > > > > + > > > > err = umount(TDIR); > > > > if (!ASSERT_OK(err, "umount " TDIR)) > > > > goto out;
On Thu, Oct 26, 2023 at 02:35:11PM -0700, Alexei Starovoitov wrote: > On Tue, Oct 24, 2023 at 10:42 PM Manu Bretelle <chantr4@gmail.com> wrote: > > > > On Tue, Oct 24, 2023 at 02:29:19PM -0700, Kui-Feng Lee wrote: > > > > > > > > > On 10/24/23 13:18, Manu Bretelle wrote: > > > > Currently this tests tries to umount /sys/kernel/debug (TDIR) but the > > > > system it is running on may have mounts below. > > > > > > > > For example, danobi/vmtest [0] VMs have > > > > mount -t tracefs tracefs /sys/kernel/debug/tracing > > > > as part of their init. > > > > > > > > This change list mounts and will umount any mounts below TDIR before > > > > umounting TDIR itself. > > > > > > > > Note that it is not umounting recursively, so in the case of a sub-mount > > > > of TDIR having another sub-mount, this will fail as mtab is ordered. > > > > > > Should we move TID to a random path likes "/sys/kernel/debug-<pid>/"? > > > > > > > Fair point, I suppose we would want to keep TDIR a defined string as it does > > simplify the gymnastic involved through the rest of the script, but yeah > > looking at the original commit: > > edb65ee5aa25 (selftests/bpf: Add bpffs preload test) > > > > I don't see any reason to use an alternate directory and rather mkdir it vs > > umounting the original one. > > so something like > > > > #define TDIR "/sys/kernel/test_bpffs" > > > > Would probably do. > > > > Alexei could confirm his original intent probably. > > I don't remember why I picked /sys/kernel/debug back then. > I suspect TDIR /tmp/foo and mkdir would work the same way. Yeah. I suspect the reason you used an existing directory is that /sys/kernel is not mutable from userspace. I ended up picking a random name under /tmp and mkdir.
diff --git a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c index 214d9f4a94a5..001bf694c269 100644 --- a/tools/testing/selftests/bpf/prog_tests/test_bpffs.c +++ b/tools/testing/selftests/bpf/prog_tests/test_bpffs.c @@ -3,12 +3,14 @@ #define _GNU_SOURCE #include <stdio.h> #include <sched.h> +#include <mntent.h> #include <sys/mount.h> #include <sys/stat.h> #include <sys/types.h> #include <test_progs.h> #define TDIR "/sys/kernel/debug" +#define MTAB "/etc/mtab" static int read_iter(char *file) { @@ -32,6 +34,8 @@ static int read_iter(char *file) static int fn(void) { + /* A buffer to store logging messages */ + char buf[1024]; struct stat a, b, c; int err, map; @@ -43,6 +47,30 @@ static int fn(void) if (!ASSERT_OK(err, "mount /")) goto out; + /* TDIR may have mounts below. unount them first */ + FILE *mtab = setmntent(MTAB, "r"); + + if (!ASSERT_TRUE(mtab != NULL, "accessing " MTAB)) { + err = errno; + goto out; + } + + struct mntent *mnt = NULL; + + while ((mnt = getmntent(mtab)) != NULL) { + if (strlen(mnt->mnt_dir) > strlen(TDIR) && + strncmp(TDIR, mnt->mnt_dir, strlen(TDIR)) == 0) { + snprintf(buf, sizeof(buf) - 1, "umount %s", mnt->mnt_dir); + err = umount(mnt->mnt_dir); + if (!ASSERT_OK(err, buf)) { + endmntent(mtab); + goto out; + } + } + } + // Ignore any error here + endmntent(mtab); + err = umount(TDIR); if (!ASSERT_OK(err, "umount " TDIR)) goto out;
Currently this tests tries to umount /sys/kernel/debug (TDIR) but the system it is running on may have mounts below. For example, danobi/vmtest [0] VMs have mount -t tracefs tracefs /sys/kernel/debug/tracing as part of their init. This change list mounts and will umount any mounts below TDIR before umounting TDIR itself. Note that it is not umounting recursively, so in the case of a sub-mount of TDIR having another sub-mount, this will fail as mtab is ordered. Test: Originally: $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" => bzImage ===> Booting ===> Setting up VM ===> Running command [ 2.138818] bpf_testmod: loading out-of-tree module taints kernel. [ 2.140913] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel bpf_testmod.ko is already unloaded. Loading bpf_testmod.ko... Successfully loaded bpf_testmod.ko. test_test_bpffs:PASS:clone 0 nsec fn:PASS:unshare 0 nsec fn:PASS:mount / 0 nsec fn:FAIL:umount /sys/kernel/debug unexpected error: -1 (errno 16) bpf_testmod.ko is already unloaded. Loading bpf_testmod.ko... Successfully loaded bpf_testmod.ko. test_test_bpffs:PASS:clone 0 nsec test_test_bpffs:PASS:waitpid 0 nsec test_test_bpffs:FAIL:bpffs test failed 255#282 test_bpffs:FAIL Summary: 0/0 PASSED, 0 SKIPPED, 1 FAILED Successfully unloaded bpf_testmod.ko. Command failed with exit code: 1 After this change: $ vmtest -k $KERNEL_REPO/arch/x86_64/boot/bzImage "./test_progs -vv -a test_bpffs" => bzImage ===> Booting ===> Setting up VM ===> Running command [ 2.035210] bpf_testmod: loading out-of-tree module taints kernel. [ 2.036510] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel bpf_testmod.ko is already unloaded. Loading bpf_testmod.ko... Successfully loaded bpf_testmod.ko. test_test_bpffs:PASS:clone 0 nsec fn:PASS:unshare 0 nsec fn:PASS:mount / 0 nsec fn:PASS:accessing /etc/mtab 0 nsec fn:PASS:umount /sys/kernel/debug/tracing 0 nsec fn:PASS:umount /sys/kernel/debug 0 nsec fn:PASS:mount tmpfs 0 nsec fn:PASS:mkdir /sys/kernel/debug/fs1 0 nsec fn:PASS:mkdir /sys/kernel/debug/fs2 0 nsec fn:PASS:mount bpffs /sys/kernel/debug/fs1 0 nsec fn:PASS:mount bpffs /sys/kernel/debug/fs2 0 nsec fn:PASS:reading /sys/kernel/debug/fs1/maps.debug 0 nsec fn:PASS:reading /sys/kernel/debug/fs2/progs.debug 0 nsec fn:PASS:creating /sys/kernel/debug/fs1/a 0 nsec fn:PASS:creating /sys/kernel/debug/fs1/a/1 0 nsec fn:PASS:creating /sys/kernel/debug/fs1/b 0 nsec fn:PASS:create_map(ARRAY) 0 nsec fn:PASS:pin map 0 nsec fn:PASS:stat(/sys/kernel/debug/fs1/a) 0 nsec fn:PASS:renameat2(/fs1/a, /fs1/b, RENAME_EXCHANGE) 0 nsec fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec fn:PASS:b should have a's inode 0 nsec fn:PASS:access(/sys/kernel/debug/fs1/b/1) 0 nsec fn:PASS:stat(/sys/kernel/debug/fs1/map) 0 nsec fn:PASS:renameat2(/fs1/c, /fs1/b, RENAME_EXCHANGE) 0 nsec fn:PASS:stat(/sys/kernel/debug/fs1/b) 0 nsec fn:PASS:b should have c's inode 0 nsec fn:PASS:access(/sys/kernel/debug/fs1/c/1) 0 nsec fn:PASS:renameat2(RENAME_NOREPLACE) 0 nsec fn:PASS:access(/sys/kernel/debug/fs1/b) 0 nsec bpf_testmod.ko is already unloaded. Loading bpf_testmod.ko... Successfully loaded bpf_testmod.ko. test_test_bpffs:PASS:clone 0 nsec test_test_bpffs:PASS:waitpid 0 nsec test_test_bpffs:PASS:bpffs test 0 nsec #282 test_bpffs:OK Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED Successfully unloaded bpf_testmod.ko. [0] https://github.com/danobi/vmtest Signed-off-by: Manu Bretelle <chantr4@gmail.com> --- .../selftests/bpf/prog_tests/test_bpffs.c | 28 +++++++++++++++++++ 1 file changed, 28 insertions(+)