mkfs01_ext3_sh / mkfs01_ext4_sh / nm01_sh from commands test in ubuntu_ltp fails

[Sean Feole]

Series: E
Problem: commands test in ubuntu_ltp fails on E, test cases listed below:
 * mkfs01_ext3_sh
 * mkfs01_ext4_sh
 * nm01_sh

 startup='Fri Dec 6 02:30:45 2019'
 mkfs01 1 TINFO: timeout per run is 0h 5m 0s
 tst_device.c:238: INFO: Using test device LTP_DEV='/dev/loop2'
 mkfs01 1 TPASS: 'mkfs -t ext3 /dev/loop2 ' passed.
 mkfs01 2 TFAIL: 'mkfs -t ext3 /dev/loop2 16000' failed, not expected.
 mkfs01 3 TPASS: 'mkfs -t ext3 -c /dev/loop2 ' passed.
 mkfs01 4 TPASS: 'mkfs -V ' passed.
 mkfs01 5 TPASS: 'mkfs -h ' passed.
 mkfs01 6 TINFO: AppArmor enabled, this may affect test results
 mkfs01 6 TINFO: it can be disabled with TST_DISABLE_APPARMOR=1 (requires super/root)
 mkfs01 6 TINFO: loaded AppArmor profiles: none

 Summary:
 passed 4
 failed 1
 skipped 0
 warnings 0
 tag=mkfs01_ext3_sh stime=1575599445 dur=2 exit=exited stat=1 core=no cu=8 cs=12

 tag=mkfs01_ext3_sh stime=1575599445 dur=2 exit=exited stat=1 core=no cu=8 cs=12
 startup='Fri Dec 6 02:30:52 2019'
 mkfs01 1 TINFO: timeout per run is 0h 5m 0s
 tst_device.c:238: INFO: Using test device LTP_DEV='/dev/loop2'
 mkfs01 1 TPASS: 'mkfs -t ext4 /dev/loop2 ' passed.
 mkfs01 2 TFAIL: 'mkfs -t ext4 /dev/loop2 16000' failed, not expected.
 mkfs01 3 TPASS: 'mkfs -t ext4 -c /dev/loop2 ' passed.
 mkfs01 4 TPASS: 'mkfs -V ' passed.
 mkfs01 5 TPASS: 'mkfs -h ' passed.
 mkfs01 6 TINFO: AppArmor enabled, this may affect test results
 mkfs01 6 TINFO: it can be disabled with TST_DISABLE_APPARMOR=1 (requires super/root)
 mkfs01 6 TINFO: loaded AppArmor profiles: none

 Summary:
 passed 4
 failed 1
 skipped 0
 warnings 0
 tag=mkfs01_ext4_sh stime=1575599452 dur=2 exit=exited stat=1 core=no cu=4 cs=1

 startup='Fri Dec 6 02:28:44 2019'
 nm01 1 TINFO: timeout per run is 0h 5m 0s
 nm01 1 TPASS: nm -f posix -A lib.a > nm.out passed as expected
 nm01 1 TPASS: Got correct listing
 nm01 1 TPASS: nm -f posix -A dir/lib.a > nm.out passed as expected
 nm01 1 TPASS: Got correct listing
 nm01 2 TPASS: nm -f posix -g /opt/ltp/testcases/data/nm01/f1 > nm.out passed as expected
 nm01 2 TPASS: Got only external symbols with -g
 nm01 3 TPASS: nm -f posix -t o /opt/ltp/testcases/data/nm01/f1 > nm.out passed as expected
 nm01 3 TPASS: Got an octal symbol values with -f
 nm01 4 TPASS: nm -f sysv /opt/ltp/testcases/data/nm01/f1 > nm.out passed as expected
 nm01 4 TPASS: Got SysV format with -f sysv
 nm01 5 TPASS: nm -f bsd /opt/ltp/testcases/data/nm01/f1 > nm_bsd.out passed as expected
 nm01 5 TPASS: nm -f posix /opt/ltp/testcases/data/nm01/f1 > nm_posix.out passed as expected
 nm01 5 TFAIL: Got wrong format with -f bsd
 0000000000003dc8 d _DYNAMIC
 0000000000003fb8 d _GLOBAL_OFFSET_TABLE_
 0000000000002000 R _IO_stdin_used
 w _ITM_deregisterTMCloneTable
 w _ITM_registerTMCloneTable
 0000000000002154 r __FRAME_END__
 0000000000002010 r __GNU_EH_FRAME_HDR
 0000000000004010 D __TMC_END__
 0000000000004010 B __bss_start
 w __cxa_finalize@@GLIBC_2.2.5
 0000000000004000 D __data_start
 0000000000001120 t __do_global_dtors_aux
 0000000000003dc0 d __do_global_dtors_aux_fini_array_entry
 0000000000004008 D __dso_handle
 0000000000003db8 d __frame_dummy_init_array_entry
 w __gmon_start__
 0000000000003dc0 d __init_array_end
 0000000000003db8 d __init_array_start
 00000000000011e0 T __libc_csu_fini
 0000000000001170 T __libc_csu_init
 U __libc_start_main@@GLIBC_2.2.5
 0000000000004010 D _edata
 0000000000004018 B _end
 00000000000011e8 T _fini
 0000000000001000 t _init
 0000000000001080 T _start
 0000000000004010 b completed.8055
 0000000000004000 W data_start
 00000000000010b0 t deregister_tm_clones
 0000000000001160 t frame_dummy
 0000000000001060 T main
 U puts@@GLIBC_2.2.5
 00000000000010e0 t register_tm_clones
 nm01 6 TPASS: nm -f sysv -u /opt/ltp/testcases/data/nm01/f1 > nm.out passed as expected
 nm01 6 TPASS: Got undefined symbols with -u
 nm01 7 TPASS: nm -s /opt/ltp/testcases/data/nm01/lib.a > nm.out passed as expected
 nm01 7 TPASS: Got index with -s
 nm01 8 TINFO: AppArmor enabled, this may affect test results
 nm01 8 TINFO: it can be disabled with TST_DISABLE_APPARMOR=1 (requires super/root)
 nm01 8 TINFO: loaded AppArmor profiles: none

 Summary:
 passed 16
 failed 1
 skipped 0
 warnings 0
 tag=nm01_sh stime=1575599324 dur=0 exit=exited stat=1 core=no cu=3 cs=0

[Po-Hsu Lin (cypressyew)]

Found on Azure 5.3.0-1009.10

[Po-Hsu Lin (cypressyew)]

Found on AWS Eoan 5.3.0-1009.10-aws

[Po-Hsu Lin (cypressyew)]

This issue CANNOT be reproduced with Eoan 5.3 generic (5.3.0-25.27) on bare metals (and even passed on s390x KVM)

[Po-Hsu Lin (cypressyew)]

Issue found on node amaura (5.3.0-29-generic)
mkfs01_ext3_sh, mkfs01_ext4_sh failed.
nm01_sh passed.

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This bug is missing log files that will aid in diagnosing the problem. While running an Ubuntu kernel (not a mainline or third-party kernel) please enter the following command in a terminal window:

apport-collect 1853610

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[Sean Feole (sfeole)]

ARM64, Focal 5.4 AWS
5.4.0-1021.21-aws

07/25 09:07:09 DEBUG| utils:0153| [stdout] tag=mkfs01_ext3_sh stime=1595667930 dur=2 exit=exited stat=1 core=no cu=12 cs=29
07/25 09:07:09 DEBUG| utils:0153| [stdout] startup='Sat Jul 25 09:05:37 2020'
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 1 TINFO: timeout per run is 0h 5m 0s
07/25 09:07:09 DEBUG| utils:0153| [stdout] tst_device.c:88: INFO: Found free device 4 '/dev/loop4'
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 1 TPASS: 'mkfs -t ext4 /dev/loop4 ' passed.
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 2 TFAIL: 'mkfs -t ext4 /dev/loop4 16000' failed, not expected.
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 3 TPASS: 'mkfs -t ext4 -c /dev/loop4 ' passed.
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 4 TPASS: 'mkfs -V ' passed.
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 5 TPASS: 'mkfs -h ' passed.
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 6 TINFO: AppArmor enabled, this may affect test results
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 6 TINFO: it can be disabled with TST_DISABLE_APPARMOR=1 (requires super/root)
07/25 09:07:09 DEBUG| utils:0153| [stdout] mkfs01 6 TINFO: loaded AppArmor profiles: none
07/25 09:07:09 DEBUG| utils:0153| [stdout]
07/25 09:07:09 DEBUG| utils:0153| [stdout] Summary:
07/25 09:07:09 DEBUG| utils:0153| [stdout] passed 4
07/25 09:07:09 DEBUG| utils:0153| [stdout] failed 1
07/25 09:07:09 DEBUG| utils:0153| [stdout] skipped 0
07/25 09:07:09 DEBUG| utils:0153| [stdout] warnings 0

[Kelsey Steele (kelsey-steele)]

Spotted on Focal/linux-azure 5.4.0-1029.29

[Kelsey Steele (kelsey-steele)]

Found on Groovy lowlatency 5.8.0-31.33

[Guilherme G. Piccoli (gpiccoli)]

Found on Focal/linux-kvm, cycle sru-20210315

[Marcelo Cerri (mhcerri)]

Still happening with linux-azure 5.8.0-1027.29.

[Guilherme G. Piccoli (gpiccoli)]

Issue observed in F/KVM, cycle sru-20210412.

[Thadeu Lima de Souza Cascardo (cascardo)]

https://lists.linux.it/pipermail/ltp/2021-April/022089.html
https://lists.linux.it/pipermail/ltp/2021-April/022123.html

This has been identified on LTP upstream, but notice that we are still seeing this.

When I test it on groovy, I see:
$ df -P -B 1k /mnt | tail -n1 | awk '{print $2}'
11388
$

Which is less than the 80%.

I wonder how people expect such a small filesystem (16MB) for 2020 decade standards to not have such a large overhead.

We should look into adding a test for an absolute overhead.

Cascardo.

[Thadeu Lima de Souza Cascardo (cascardo)]

Journal size is counted as overhead and it's at least 1024 blocks, which is 4096 "1k" blocks for a filesystem with default blocksize of 4KiB. For a 16000 filesystem, this is already 20%.

Cascardo.

[Thadeu Lima de Souza Cascardo (cascardo)]

A patch we carry on e2fsprogs changes the default blocksize for small filesystems. That is what breaks the test. We should probably add the block size when creating these filesystems under LTP.

Description: Use 4k blocksize in all ext4 mke2fs.conf such that lvm migration
  between non-4k PVs and 4k PVs works irrespective of the volume size.
Author: Dimitri John Ledkov <email address hidden>
Bug-Ubuntu: https://bugs.launchpad.net/bugs/1817097

--- e2fsprogs-1.45.1.orig/misc/mke2fs.conf.in
+++ e2fsprogs-1.45.1/misc/mke2fs.conf.in
@@ -15,12 +15,10 @@
                inode_size = 256
        }
        small = {
- blocksize = 1024
                inode_size = 128
                inode_ratio = 4096
        }
        floppy = {
- blocksize = 1024
                inode_size = 128
                inode_ratio = 8192
        }

[Thadeu Lima de Souza Cascardo (cascardo)]

https://lists.linux.it/pipermail/ltp/2021-May/022536.html

[Po-Hsu Lin (cypressyew)]

This has landed in https://github.com/linux-test-project/ltp/commit/15d0ec59fcb4bee3ae2e8cbb30a868fc1441e2c8

And this failure cannot be found on:
F-aws / F-azure / H-5.11

Note: this is not tested on F-kvm

Therefore I will mark this as fix-released. Thanks.