Re: [LTP] LTP: syscalls: regression on mainline - ioctl_loop01 mknod07 setns01
Hi Yang, On Fri, Jun 5, 2020 at 11:27 AM Yang Xu wrote: > since ~LOOP_SET_STATUS_SETTABLE_FLAGS has been included in > ~LOOP_SET_STATUS_CLEARABLE_FLAGS, do we still need the previous step? > What do you think about it? Yeah, I don't think we need the previous step with the current set of flags, because clearable is a subset of settable. I will send a follow-up patch some time next week. Best, Martijn > > Best Regards > Yang Xu > > > Hey Yang, > > > > On Fri, Jun 5, 2020 at 10:59 AM Yang Xu > > wrote: > >> > >> Hi Martijn > >> > >> Sorry for noise. I see your patch in here[1] . I will modify > >> ioctl_loop01 to test that LO_FLAGS_PARTSCAN can not clear and > >> LO_FLAGS_AUTOCLEAR can be clear. > > > > Thanks, that would indeed be useful. > > > >> > >> ps: Giving the url of patch is better so that other people doesn't need > >> to investigate it again. > >> [1]https://patchwork.kernel.org/patch/11588321/ > > > > Ok, will do next time! > > > > Best, > > Martijn > >> > >> Best Regards > >> Yang Xu > >>> Hi Martijn > >>> > Hi Naresh, > > I just sent a patch and cc'd you. I verified all the loop tests pass > again with that patch. > >>> I think you want to say "without". I verified the ioctl_loop01 fails > >>> with faf1d25440 ("loop: Clean up LOOP_SET_STATUS lo_flags handling"). > >>> > >>> This kernel commit breaks old behaviour(if old flag all 0, new flag is > >>> always 0 regradless your flag setting). > >>> > >>> I think we should modify code as below: > >>> diff --git a/drivers/block/loop.c b/drivers/block/loop.c > >>> index 13518ba191f5..c6ba8cf486ce 100644 > >>> --- a/drivers/block/loop.c > >>> +++ b/drivers/block/loop.c > >>> @@ -1364,11 +1364,9 @@ loop_set_status(struct loop_device *lo, const > >>> struct loop_info64 *info) > >>> if (err) > >>> goto out_unfreeze; > >>> > >>> - /* Mask out flags that can't be set using LOOP_SET_STATUS. */ > >>> - lo->lo_flags &= ~LOOP_SET_STATUS_SETTABLE_FLAGS; > >>> - /* For those flags, use the previous values instead */ > >>> - lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS; > >>> - /* For flags that can't be cleared, use previous values too */ > >>> + /* Mask out flags that can be set using LOOP_SET_STATUS. */ > >>> + lo->lo_flags &= LOOP_SET_STATUS_SETTABLE_FLAGS; > >>> + /* For flags that can't be cleared, use previous values. */ > >>> lo->lo_flags |= prev_lo_flags &~LOOP_SET_STATUS_CLEARABLE_FLAGS; > >>> > >>> Best Regards > >>> Yang Xu > > Thanks, > Martijn > > > On Thu, Jun 4, 2020 at 9:10 PM Martijn Coenen wrote: > > > > Hi Naresh, > > > > I suspect the loop failures are due to > > faf1d25440d6ad06d509dada4b6fe62fea844370 ("loop: Clean up > > LOOP_SET_STATUS lo_flags handling"), I will investigate and get back > > to you. > > > > Thanks, > > Martijn > > > > On Thu, Jun 4, 2020 at 7:19 PM Naresh Kamboju > > wrote: > >> > >> + linux-bl...@vger.kernel.org > >> > >> On Thu, 4 Jun 2020 at 22:47, Naresh Kamboju > >> wrote: > >>> > >>> Following three test cases reported as regression on Linux mainline > >>> kernel > >>> on x86_64, arm64, arm and i386 > >>> > >>> ltp-syscalls-tests: > >>> * ioctl_loop01 > >>> * mknod07 > >>> * setns01 > >>> > >>> git repo: > >>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git > >>> git branch: master > >>> GOOD: > >>> git commit: b23c4771ff62de8ca9b5e4a2d64491b2fb6f8f69 > >>> git describe: v5.7-1230-gb23c4771ff62 > >>> BAD: > >>> git commit: 1ee08de1e234d95b5b4f866878b72fceb5372904 > >>> git describe: v5.7-3523-g1ee08de1e234 > >>> > >>> kernel-config: > >>> https://builds.tuxbuild.com/U3bU0dMA62OVHb4DvZIVuw/kernel.config > >>> > >>> We are investigating these failures. > >>> > >>> tst_test.c:906: CONF: btrfs driver not available > >>> tst_test.c:1246: INFO: Timeout per run is 0h 15m 00s > >>> tst_device.c:88: INFO: Found free device 1 '/dev/loop1' > >>> ioctl_loop01.c:49: PASS: /sys/block/loop1/loop/partscan = 0 > >>> [ 1073.639677] loop_set_status: loop1 () has still dirty pages > >>> (nrpages=1) > >>> ioctl_loop01.c:50: PASS: /sys/block/loop1/loop/autoclear = 0 > >>> ioctl_loop01.c:51: PASS: /sys/block/loop1/loop/backing_file = > >>> '/scratch/ltp-mnIdulzriQ/9cPtLQ/test.img' > >>> ioctl_loop01.c:63: FAIL: expect 12 but got 17 > >>> ioctl_loop01.c:67: FAIL: /sys/block/loop1/loop/partscan != 1 got 0 > >>> ioctl_loop01.c:68: FAIL: /sys/block/loop1/loop/autoclear != 1 got 0 > >>> ioctl_loop01.c:79: FAIL: access /dev/loop1p1 fails > >>> [ 1073.679678] loop_set_status: loop1 () has still dirty pages > >>> (nrpages=1) > >>> ioctl_loop01.c:85: FAIL: access /sys/block/loop1/loop1p1 fails >
Re: [LTP] LTP: syscalls: regression on mainline - ioctl_loop01 mknod07 setns01
Hi Martign Also for your kernel commit, lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS; lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_CLEARABLE_FLAGS; since ~LOOP_SET_STATUS_SETTABLE_FLAGS has been included in ~LOOP_SET_STATUS_CLEARABLE_FLAGS, do we still need the previous step? What do you think about it? Best Regards Yang Xu Hey Yang, On Fri, Jun 5, 2020 at 10:59 AM Yang Xu wrote: Hi Martijn Sorry for noise. I see your patch in here[1] . I will modify ioctl_loop01 to test that LO_FLAGS_PARTSCAN can not clear and LO_FLAGS_AUTOCLEAR can be clear. Thanks, that would indeed be useful. ps: Giving the url of patch is better so that other people doesn't need to investigate it again. [1]https://patchwork.kernel.org/patch/11588321/ Ok, will do next time! Best, Martijn Best Regards Yang Xu Hi Martijn Hi Naresh, I just sent a patch and cc'd you. I verified all the loop tests pass again with that patch. I think you want to say "without". I verified the ioctl_loop01 fails with faf1d25440 ("loop: Clean up LOOP_SET_STATUS lo_flags handling"). This kernel commit breaks old behaviour(if old flag all 0, new flag is always 0 regradless your flag setting). I think we should modify code as below: diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 13518ba191f5..c6ba8cf486ce 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -1364,11 +1364,9 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) if (err) goto out_unfreeze; - /* Mask out flags that can't be set using LOOP_SET_STATUS. */ - lo->lo_flags &= ~LOOP_SET_STATUS_SETTABLE_FLAGS; - /* For those flags, use the previous values instead */ - lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS; - /* For flags that can't be cleared, use previous values too */ + /* Mask out flags that can be set using LOOP_SET_STATUS. */ + lo->lo_flags &= LOOP_SET_STATUS_SETTABLE_FLAGS; + /* For flags that can't be cleared, use previous values. */ lo->lo_flags |= prev_lo_flags &~LOOP_SET_STATUS_CLEARABLE_FLAGS; Best Regards Yang Xu Thanks, Martijn On Thu, Jun 4, 2020 at 9:10 PM Martijn Coenen wrote: Hi Naresh, I suspect the loop failures are due to faf1d25440d6ad06d509dada4b6fe62fea844370 ("loop: Clean up LOOP_SET_STATUS lo_flags handling"), I will investigate and get back to you. Thanks, Martijn On Thu, Jun 4, 2020 at 7:19 PM Naresh Kamboju wrote: + linux-bl...@vger.kernel.org On Thu, 4 Jun 2020 at 22:47, Naresh Kamboju wrote: Following three test cases reported as regression on Linux mainline kernel on x86_64, arm64, arm and i386 ltp-syscalls-tests: * ioctl_loop01 * mknod07 * setns01 git repo: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git git branch: master GOOD: git commit: b23c4771ff62de8ca9b5e4a2d64491b2fb6f8f69 git describe: v5.7-1230-gb23c4771ff62 BAD: git commit: 1ee08de1e234d95b5b4f866878b72fceb5372904 git describe: v5.7-3523-g1ee08de1e234 kernel-config: https://builds.tuxbuild.com/U3bU0dMA62OVHb4DvZIVuw/kernel.config We are investigating these failures. tst_test.c:906: CONF: btrfs driver not available tst_test.c:1246: INFO: Timeout per run is 0h 15m 00s tst_device.c:88: INFO: Found free device 1 '/dev/loop1' ioctl_loop01.c:49: PASS: /sys/block/loop1/loop/partscan = 0 [ 1073.639677] loop_set_status: loop1 () has still dirty pages (nrpages=1) ioctl_loop01.c:50: PASS: /sys/block/loop1/loop/autoclear = 0 ioctl_loop01.c:51: PASS: /sys/block/loop1/loop/backing_file = '/scratch/ltp-mnIdulzriQ/9cPtLQ/test.img' ioctl_loop01.c:63: FAIL: expect 12 but got 17 ioctl_loop01.c:67: FAIL: /sys/block/loop1/loop/partscan != 1 got 0 ioctl_loop01.c:68: FAIL: /sys/block/loop1/loop/autoclear != 1 got 0 ioctl_loop01.c:79: FAIL: access /dev/loop1p1 fails [ 1073.679678] loop_set_status: loop1 () has still dirty pages (nrpages=1) ioctl_loop01.c:85: FAIL: access /sys/block/loop1/loop1p1 fails HINT: You _MAY_ be missing kernel fixes, see: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=10c70d95c0f2 mke2fs 1.43.8 (1-Jan-2018) [ 1264.711379] EXT4-fs (loop0): mounting ext2 file system using the ext4 subsystem [ 1264.716642] EXT4-fs (loop0): mounted filesystem without journal. Opts: (null) mknod07 0 TINFO : Using test device LTP_DEV='/dev/loop0' mknod07 0 TINFO : Formatting /dev/loop0 with ext2 opts='' extra opts='' mknod07 1 TPASS : mknod failed as expected: TEST_ERRNO=EACCES(13): Permission denied mknod07 2 TPASS : mknod failed as expected: TEST_ERRNO=EACCES(13): Permission denied mknod07 3 TFAIL : mknod07.c:155: mknod succeeded unexpectedly mknod07 4 TPASS : mknod failed as expected: TEST_ERRNO=EPERM(1): Operation not permitted mknod07 5 TPASS : mknod failed as expected: TEST_ERRNO=EROFS(30): Read-only file system mknod07 6 TPASS : mknod failed as
Re: [LTP] LTP: syscalls: regression on mainline - ioctl_loop01 mknod07 setns01
Hey Yang, On Fri, Jun 5, 2020 at 10:59 AM Yang Xu wrote: > > Hi Martijn > > Sorry for noise. I see your patch in here[1] . I will modify > ioctl_loop01 to test that LO_FLAGS_PARTSCAN can not clear and > LO_FLAGS_AUTOCLEAR can be clear. Thanks, that would indeed be useful. > > ps: Giving the url of patch is better so that other people doesn't need > to investigate it again. > [1]https://patchwork.kernel.org/patch/11588321/ Ok, will do next time! Best, Martijn > > Best Regards > Yang Xu > > Hi Martijn > > > >> Hi Naresh, > >> > >> I just sent a patch and cc'd you. I verified all the loop tests pass > >> again with that patch. > > I think you want to say "without". I verified the ioctl_loop01 fails > > with faf1d25440 ("loop: Clean up LOOP_SET_STATUS lo_flags handling"). > > > > This kernel commit breaks old behaviour(if old flag all 0, new flag is > > always 0 regradless your flag setting). > > > > I think we should modify code as below: > > diff --git a/drivers/block/loop.c b/drivers/block/loop.c > > index 13518ba191f5..c6ba8cf486ce 100644 > > --- a/drivers/block/loop.c > > +++ b/drivers/block/loop.c > > @@ -1364,11 +1364,9 @@ loop_set_status(struct loop_device *lo, const > > struct loop_info64 *info) > > if (err) > > goto out_unfreeze; > > > > - /* Mask out flags that can't be set using LOOP_SET_STATUS. */ > > - lo->lo_flags &= ~LOOP_SET_STATUS_SETTABLE_FLAGS; > > - /* For those flags, use the previous values instead */ > > - lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS; > > - /* For flags that can't be cleared, use previous values too */ > > + /* Mask out flags that can be set using LOOP_SET_STATUS. */ > > + lo->lo_flags &= LOOP_SET_STATUS_SETTABLE_FLAGS; > > + /* For flags that can't be cleared, use previous values. */ > > lo->lo_flags |= prev_lo_flags &~LOOP_SET_STATUS_CLEARABLE_FLAGS; > > > > Best Regards > > Yang Xu > >> > >> Thanks, > >> Martijn > >> > >> > >> On Thu, Jun 4, 2020 at 9:10 PM Martijn Coenen wrote: > >>> > >>> Hi Naresh, > >>> > >>> I suspect the loop failures are due to > >>> faf1d25440d6ad06d509dada4b6fe62fea844370 ("loop: Clean up > >>> LOOP_SET_STATUS lo_flags handling"), I will investigate and get back > >>> to you. > >>> > >>> Thanks, > >>> Martijn > >>> > >>> On Thu, Jun 4, 2020 at 7:19 PM Naresh Kamboju > >>> wrote: > > + linux-bl...@vger.kernel.org > > On Thu, 4 Jun 2020 at 22:47, Naresh Kamboju > wrote: > > > > Following three test cases reported as regression on Linux mainline > > kernel > > on x86_64, arm64, arm and i386 > > > >ltp-syscalls-tests: > > * ioctl_loop01 > > * mknod07 > > * setns01 > > > > git repo: > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git > > git branch: master > > GOOD: > >git commit: b23c4771ff62de8ca9b5e4a2d64491b2fb6f8f69 > >git describe: v5.7-1230-gb23c4771ff62 > > BAD: > >git commit: 1ee08de1e234d95b5b4f866878b72fceb5372904 > >git describe: v5.7-3523-g1ee08de1e234 > > > > kernel-config: > > https://builds.tuxbuild.com/U3bU0dMA62OVHb4DvZIVuw/kernel.config > > > > We are investigating these failures. > > > > tst_test.c:906: CONF: btrfs driver not available > > tst_test.c:1246: INFO: Timeout per run is 0h 15m 00s > > tst_device.c:88: INFO: Found free device 1 '/dev/loop1' > > ioctl_loop01.c:49: PASS: /sys/block/loop1/loop/partscan = 0 > > [ 1073.639677] loop_set_status: loop1 () has still dirty pages > > (nrpages=1) > > ioctl_loop01.c:50: PASS: /sys/block/loop1/loop/autoclear = 0 > > ioctl_loop01.c:51: PASS: /sys/block/loop1/loop/backing_file = > > '/scratch/ltp-mnIdulzriQ/9cPtLQ/test.img' > > ioctl_loop01.c:63: FAIL: expect 12 but got 17 > > ioctl_loop01.c:67: FAIL: /sys/block/loop1/loop/partscan != 1 got 0 > > ioctl_loop01.c:68: FAIL: /sys/block/loop1/loop/autoclear != 1 got 0 > > ioctl_loop01.c:79: FAIL: access /dev/loop1p1 fails > > [ 1073.679678] loop_set_status: loop1 () has still dirty pages > > (nrpages=1) > > ioctl_loop01.c:85: FAIL: access /sys/block/loop1/loop1p1 fails > > > > HINT: You _MAY_ be missing kernel fixes, see: > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=10c70d95c0f2 > > > > > > mke2fs 1.43.8 (1-Jan-2018) > > [ 1264.711379] EXT4-fs (loop0): mounting ext2 file system using the > > ext4 subsystem > > [ 1264.716642] EXT4-fs (loop0): mounted filesystem without journal. > > Opts: (null) > > mknod07 0 TINFO : Using test device LTP_DEV='/dev/loop0' > > mknod07 0 TINFO : Formatting /dev/loop0 with ext2 opts='' > > extra opts='' > > mknod07 1 TPASS : mknod failed as expected: > > TEST_ERRNO=EACCES(13): Permission denied > > mknod07 2 TPASS : mknod failed as
Re: [LTP] LTP: syscalls: regression on mainline - ioctl_loop01 mknod07 setns01
Hi Martijn Sorry for noise. I see your patch in here[1] . I will modify ioctl_loop01 to test that LO_FLAGS_PARTSCAN can not clear and LO_FLAGS_AUTOCLEAR can be clear. ps: Giving the url of patch is better so that other people doesn't need to investigate it again. [1]https://patchwork.kernel.org/patch/11588321/ Best Regards Yang Xu Hi Martijn Hi Naresh, I just sent a patch and cc'd you. I verified all the loop tests pass again with that patch. I think you want to say "without". I verified the ioctl_loop01 fails with faf1d25440 ("loop: Clean up LOOP_SET_STATUS lo_flags handling"). This kernel commit breaks old behaviour(if old flag all 0, new flag is always 0 regradless your flag setting). I think we should modify code as below: diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 13518ba191f5..c6ba8cf486ce 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -1364,11 +1364,9 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info) if (err) goto out_unfreeze; - /* Mask out flags that can't be set using LOOP_SET_STATUS. */ - lo->lo_flags &= ~LOOP_SET_STATUS_SETTABLE_FLAGS; - /* For those flags, use the previous values instead */ - lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS; - /* For flags that can't be cleared, use previous values too */ + /* Mask out flags that can be set using LOOP_SET_STATUS. */ + lo->lo_flags &= LOOP_SET_STATUS_SETTABLE_FLAGS; + /* For flags that can't be cleared, use previous values. */ lo->lo_flags |= prev_lo_flags &~LOOP_SET_STATUS_CLEARABLE_FLAGS; Best Regards Yang Xu Thanks, Martijn On Thu, Jun 4, 2020 at 9:10 PM Martijn Coenen wrote: Hi Naresh, I suspect the loop failures are due to faf1d25440d6ad06d509dada4b6fe62fea844370 ("loop: Clean up LOOP_SET_STATUS lo_flags handling"), I will investigate and get back to you. Thanks, Martijn On Thu, Jun 4, 2020 at 7:19 PM Naresh Kamboju wrote: + linux-bl...@vger.kernel.org On Thu, 4 Jun 2020 at 22:47, Naresh Kamboju wrote: Following three test cases reported as regression on Linux mainline kernel on x86_64, arm64, arm and i386 ltp-syscalls-tests: * ioctl_loop01 * mknod07 * setns01 git repo: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git git branch: master GOOD: git commit: b23c4771ff62de8ca9b5e4a2d64491b2fb6f8f69 git describe: v5.7-1230-gb23c4771ff62 BAD: git commit: 1ee08de1e234d95b5b4f866878b72fceb5372904 git describe: v5.7-3523-g1ee08de1e234 kernel-config: https://builds.tuxbuild.com/U3bU0dMA62OVHb4DvZIVuw/kernel.config We are investigating these failures. tst_test.c:906: CONF: btrfs driver not available tst_test.c:1246: INFO: Timeout per run is 0h 15m 00s tst_device.c:88: INFO: Found free device 1 '/dev/loop1' ioctl_loop01.c:49: PASS: /sys/block/loop1/loop/partscan = 0 [ 1073.639677] loop_set_status: loop1 () has still dirty pages (nrpages=1) ioctl_loop01.c:50: PASS: /sys/block/loop1/loop/autoclear = 0 ioctl_loop01.c:51: PASS: /sys/block/loop1/loop/backing_file = '/scratch/ltp-mnIdulzriQ/9cPtLQ/test.img' ioctl_loop01.c:63: FAIL: expect 12 but got 17 ioctl_loop01.c:67: FAIL: /sys/block/loop1/loop/partscan != 1 got 0 ioctl_loop01.c:68: FAIL: /sys/block/loop1/loop/autoclear != 1 got 0 ioctl_loop01.c:79: FAIL: access /dev/loop1p1 fails [ 1073.679678] loop_set_status: loop1 () has still dirty pages (nrpages=1) ioctl_loop01.c:85: FAIL: access /sys/block/loop1/loop1p1 fails HINT: You _MAY_ be missing kernel fixes, see: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=10c70d95c0f2 mke2fs 1.43.8 (1-Jan-2018) [ 1264.711379] EXT4-fs (loop0): mounting ext2 file system using the ext4 subsystem [ 1264.716642] EXT4-fs (loop0): mounted filesystem without journal. Opts: (null) mknod07 0 TINFO : Using test device LTP_DEV='/dev/loop0' mknod07 0 TINFO : Formatting /dev/loop0 with ext2 opts='' extra opts='' mknod07 1 TPASS : mknod failed as expected: TEST_ERRNO=EACCES(13): Permission denied mknod07 2 TPASS : mknod failed as expected: TEST_ERRNO=EACCES(13): Permission denied mknod07 3 TFAIL : mknod07.c:155: mknod succeeded unexpectedly mknod07 4 TPASS : mknod failed as expected: TEST_ERRNO=EPERM(1): Operation not permitted mknod07 5 TPASS : mknod failed as expected: TEST_ERRNO=EROFS(30): Read-only file system mknod07 6 TPASS : mknod failed as expected: TEST_ERRNO=ELOOP(40): Too many levels of symbolic links setns01 0 TINFO : ns_name=ipc, ns_fds[0]=6, ns_types[0]=0x800 setns01 0 TINFO : ns_name=mnt, ns_fds[1]=7, ns_types[1]=0x2 setns01 0 TINFO : ns_name=net, ns_fds[2]=8, ns_types[2]=0x4000 setns01 0 TINFO : ns_name=pid, ns_fds[3]=9, ns_types[3]=0x2000 setns01 0 TINFO : ns_name=uts, ns_fds[4]=10, ns_types[4]=0x400 setns01 0 TINFO : setns(-1, 0x800)