Skip to contentSkip to conversation

e2fsprogs - possible regression between 1.42.7 and 1.42.8

DISCUSSION Nikola Ciprich 3 participants 9 messages Jul 29, 2013 – Sep 4, 2013 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

Hi,

trying to compile latest e2fsprogs and running check under RHEL6-compatible distro (centos) fails:

r_1024_small_bg: ext2 1024 blocksize with small block groups: failed
r_ext4_small_bg: ext4 1024 blocksize with small block groups: failed

dunno whether this is known issue... will bisect help?

with regards

nik

Signature · 13 lines
-------------------------------------
Ing. Nikola CIPRICH
LinuxBox.cz, s.r.o.
28.rijna 168, 709 00 Ostrava

tel.:   +420 591 166 214
fax:    +420 596 621 273
mobil:  +420 777 093 799
www.linuxbox.cz

mobil servis: +420 737 238 656
email servis: [email protected]
-------------------------------------
#1 lore raw
Quoted 8 lines · Nikola Ciprich

trying to compile latest e2fsprogs and running check under RHEL6-compatible distro (centos) fails:

r_1024_small_bg: ext2 1024 blocksize with small block groups: failed
r_ext4_small_bg: ext4 1024 blocksize with small block groups: failed

dunno whether this is known issue... will bisect help?

What platform are you running this on? X86? I generally do a full regression test run before I do an e2fsprogs release, so it may very well be something which might be platform or compiler specific.

Can you send me the r_1024_small_bg.failed and r_ext4_small_bg.failed files?

Thanks,

- Ted

#2 lore raw

Hello,

Quoted 3 lines · Theodore Ts'o

What platform are you running this on? X86? I generally do a full regression test run before I do an e2fsprogs release, so it may very well be something which might be platform or compiler specific.

it's x86_64 running kernel 3.0.88-rc1. I also tried x86_64 centos 5, and different test fails (I was successfully compiling previous releases on those boxes).

Quoted 3 lines

Can you send me the r_1024_small_bg.failed and r_ext4_small_bg.failed files?

sure, although both those files are empty.. I've uploaded test logs (including rhel5 failed one) here:

http://nik.lbox.cz/download/e2fsprogs-tests/

please let me know if I could provide more information

BR

nik

Quoted 5 lines

Thanks,

- Ted

Signature · 14 lines
-------------------------------------
Ing. Nikola CIPRICH
LinuxBox.cz, s.r.o.
28. rijna 168, 709 00 Ostrava

tel.:   +420 591 166 214
fax:    +420 596 621 273
mobil:  +420 777 093 799

www.linuxbox.cz

mobil servis: +420 737 238 656
email servis: [email protected]
-------------------------------------
#3 lore raw
In reply to Nikola Ciprich

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 5 lines · Nikola Ciprich

Hi,

trying to compile latest e2fsprogs and running check under RHEL6-compatible distro (centos) fails:

Quoted 9 lines

r_1024_small_bg: ext2 1024 blocksize with small block groups: failed
r_ext4_small_bg: ext4 1024 blocksize with small block groups: failed

dunno whether this is known issue... will bisect help?

with regards

nik

Ted, these are the same ones I saw, plus one I think (working on getting all the info).

I don't think it's a regression, because:

commit e79a9395b382e831c125d000d2bf16ba4b6253d4

Author: Theodore Ts'o <[email protected]>
Date:   Sun Mar 31 20:34:24 2013 -0400

tests: add more tests for off-line resizing

and:

$ git describe --contains e79a9395b382e831c125d000d2bf16ba4b6253d4
v1.42.8~31

the tests were only added in the last release. Running the same tests on older releases most likely breaks as well.

- -Eric
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
Comment: GPGTools - http://gpgtools.org
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQIcBAEBAgAGBQJR9pS1AAoJECCuFpLhPd7g+dMP/1MdRJN8AzVIL6iX8sQ/yo2F atDLdkzDjyjS+UzsTPEP0Q7C9MjSl74Z7+57SNyKjgESAY1MOp8LlPSsFMiuJfbW u7Myd9iWfVHK/DIlOPBvW55WAi8y7w6HhXcTKhbjuEPUPJAPb1H36p+iwb/Sd40H A/Ti6yOYyI05rXHonG+zzD4TFG3gNQT0zg3O0+BQZgX5uFuENlehgoZXxH61dwtP tQTbFCrjcn337fTukwKDgDSBzuQT1Z6cQemzmoT+2C9Nhkqztjm6MMX+xjd96W0R /aPITdVC4zLVl15QZSv/knAP0fJJyhhKsGJOYvzGXW2ydhFu/EmLtWSvu80E+slf tigSqmh98WJzqyrrxcLOmV/ZncYKke/hpWa20t3eD775Lx4znkOGt7XkB7RtK648 LAcTrGi6Z7Saj44kS2zOvWs+IcxEzKAZWpZ2tFxxa5kn1riuHimhIlQZJwIAVF+j XJT4c0IMGOIan0Z0NuP+K68RVi6X6Rsvr+o5OiUv6N4pnv8DsIskDAmpYpRUSwPV R5WlwidCOpnPoiff8dAOnFjDZnmxG0VUMq9URTtvchNwPflBgp6LDC7LLbb+HWkh uxCOqiirv1T1ZTGIglN804CzccWHjTufNZz6p/5klApxBa/HUZuzb29YQ7oicUSy 3fiWWRobTPA/+PQGKGBT
=G+1+
-----END PGP SIGNATURE-----

#4 lore raw
Quoted 12 lines · Eric Sandeen

Ted, these are the same ones I saw, plus one I think (working on getting all the info).

I don't think it's a regression, because:

commit e79a9395b382e831c125d000d2bf16ba4b6253d4

Author: Theodore Ts'o <[email protected]>
Date:   Sun Mar 31 20:34:24 2013 -0400

tests: add more tests for off-line resizing

Quoted 7 lines

and:

$ git describe --contains e79a9395b382e831c125d000d2bf16ba4b6253d4
v1.42.8~31

the tests were only added in the last release. Running the same tests on older releases most likely breaks as well.

Well, they would almost certainly break on older releases because of bugs due to bugs that were fixed in 1.42.8. :-)

Let me be more precise; these tests aren't failing for me when I run build and run "make check" on pristine 1.42.8 version of e2fsprogs on Debian Stable. So it's likely that the test is doing something that is specific to Red Hat systems. It may stil be turning up a bug that for some reason we're not seeing on Debian systems.

We are using the e2fsck binary built in the tree as the source of test bits for the resize test. I'm guessing that it is substantially smaller or bigger when built on Red Hat systems?!?

Could you modify tests/script/resize to capture a copy of the constructed file system before we start running resize2fs on it so I can try reproducing it on my end?

- Ted

P.S. Hmm, for some reason the size of the e2fsck binary must be *substantially* smaller on Red Hat systems. What configure options are you using?

Looking at my log, it shrinks the file system to:

r_1024_small_bg.log:The filesystem on /tmp/e2fsprogs-tmp.RE74xl is now 1341 blocks long.

and then

r_1024_small_bg.log:The filesystem on /tmp/e2fsprogs-tmp.RE74xl is now 1215 blocks long.

On your log, it shrinks the filesystem to 1191 blocks and then 1111 blocks.

On my system with default configure options, the size of the e2fsck binary is 1124k. It sounds like the size of your compiled e2fsck binary is approximately 100k smaller?

#5 lore raw
Quoted 15 lines · Theodore Ts'o
Quoted 14 lines · Eric Sandeen

Ted, these are the same ones I saw, plus one I think (working on getting all the info).

I don't think it's a regression, because:

commit e79a9395b382e831c125d000d2bf16ba4b6253d4

Author: Theodore Ts'o <[email protected]>
Date:   Sun Mar 31 20:34:24 2013 -0400

tests: add more tests for off-line resizing

and:

Quoted 8 lines
Quoted 5 lines
$ git describe --contains e79a9395b382e831c125d000d2bf16ba4b6253d4
v1.42.8~31

the tests were only added in the last release. Running the same tests on older releases most likely breaks as well.

Well, they would almost certainly break on older releases because of bugs due to bugs that were fixed in 1.42.8. :-)

...or not ;)

Quoted 9 lines

Let me be more precise; these tests aren't failing for me when I run build and run "make check" on pristine 1.42.8 version of e2fsprogs on Debian Stable. So it's likely that the test is doing something that is specific to Red Hat systems. It may stil be turning up a bug that for some reason we're not seeing on Debian systems.

We are using the e2fsck binary built in the tree as the source of test bits for the resize test. I'm guessing that it is substantially smaller or bigger when built on Red Hat systems?!?

Something along those lines.

Quoted 3 lines

Could you modify tests/script/resize to capture a copy of the constructed file system before we start running resize2fs on it so I can try reproducing it on my end?

Yes, getting to it ...

Quoted 5 lines

- Ted

P.S. Hmm, for some reason the size of the e2fsck binary must be *substantially* smaller on Red Hat systems. What configure options are you using?

oh we just symlink it to /bin/true ;) (KIDDING)

Hm, it's currently this in the specfile:

%configure --enable-elf-shlibs --enable-nls --disable-uuidd --disable-fsck \
           --disable-e2initrd-helper --disable-libblkid --disable-libuuid \
           --with-root-prefix=/usr

and %configure pulls in other stuff as well - let's see, here's the full cmdline:

+ ./configure --build=s390x-redhat-linux-gnu --host=s390x-redhat-linux-gnu --program-prefix= --disable-dependency-tracking --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --sysconfdir=/etc --datadir=/usr/share --includedir=/usr/include --libdir=/usr/lib64 --libexecdir=/usr/libexec --localstatedir=/var --sharedstatedir=/var/lib --mandir=/usr/share/man --infodir=/usr/share/info --enable-elf-shlibs --enable-nls --disable-uuidd --disable-fsck --disable-e2initrd-helper --disable-libblkid --disable-libuuid --with-root-prefix=/usr

(!)

-Eric

Quoted 14 lines

Looking at my log, it shrinks the file system to:

r_1024_small_bg.log:The filesystem on /tmp/e2fsprogs-tmp.RE74xl is now 1341 blocks long.

and then

r_1024_small_bg.log:The filesystem on /tmp/e2fsprogs-tmp.RE74xl is now 1215 blocks long.

On your log, it shrinks the filesystem to 1191 blocks and then 1111 blocks.

On my system with default configure options, the size of the e2fsck binary is 1124k. It sounds like the size of your compiled e2fsck binary is approximately 100k smaller?

A message that is not in the archive (sent off-list or not yet received): [email protected]

#7 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 14 lines · Eric Sandeen
Quoted 12 lines · Eric Sandeen
Quoted 11 lines · Nikola Ciprich

Hi,

trying to compile latest e2fsprogs and running check under RHEL6-compatible distro (centos) fails:

r_1024_small_bg: ext2 1024 blocksize with small block groups: failed
r_ext4_small_bg: ext4 1024 blocksize with small block groups: failed

dunno whether this is known issue... will bisect help?

with regards

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

nik

Ted, these are the same ones I saw, plus one I think (working on getting all the info).

Sorry this took a while. Attached is a qcow image of a broken r_1024_small_bg filesystem. Doing resize2fs -M on it twice should corrupt it, even on x86_64.

Sorry - the image as attached is not broken, but 2 current resize2fs -M's break it.

- -Eric

Quoted 8 lines

(aside: the test is a bit weird, it does:

echo $RESIZE2FS $RESIZE2FS_OPTS -d 31 -M $TMPFILE $SIZE_2 >> $LOG 2>&1

but specifying -M as well as a size doesn't make much sense?)

-Eric

-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
Comment: GPGTools - http://gpgtools.org
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQIcBAEBAgAGBQJR+HIsAAoJECCuFpLhPd7gT3cP/3YLhpclAhU6UHIW4cqyKhIl STIwV3ROZmic5qQb8X6l5etc8ALCs2b5LVYjhlMlZWISkN64pXOhL0lz7coSABM9 wnZU2y3LntcyGFEg7aEs+sXe2mXqX139NkoNwKfBBPidQ/HdwpnnTdlu4FZeTyrj aDHIUGPs2YBbp2QaBhmzl5b7xwEoHCq/XtrH7v7+SHyHEGzIYTn3w96vyXqVYTjs csUgeMhFFw58Hq+/03UE13Ig/Et8AqSgonnaEiG3ZKXsDIxhjFxqQPGsdD2BleGd Rb/fxnkM0RKhr7xSK3CFWnLQ8LkaUtexazxvr+7BmmgRx8jyQhQj8ASYXzmUuh/S RtbnZ3wioxEwCRhr3FP0ZpoOZwbkDWd/UBy407ZBblxVafjLSYbsMHDM7rPrQHoz oSo1W4c7sy85pNfCgMnQYyuqik5KVDuQNk/nqEkv6tE1YOb5zXTrpDKjEFgBKYVd FpLRk9f0rhoxcwyfZsIY6Wx/4h4rWKnhK6jqL5unDLJitPv405KlNfY28AoFNme4 ZxwDSddausNm+8u9sZON7cuyhAb70Q8N0W6ECPX8+ie8fvXwPDRSJYPpMXqOBK06 35SPqxKlCpT7BGMAqr7+h8qemYdcTS0z7ag2im+MRS1t58AFSsuIyj+IirPWwTUM SQRtuG3mOsWE7l93c8bc
=96jm
-----END PGP SIGNATURE-----

#8 lore raw

Hello Eric,

I'm very sorry for such a late reply, I was a bit busy with other projects.. Well, I don't really understand what I should do with the images now :) But if I understand previous discussion well, my problem is not caused by code regressions, but new tests which were added to this new version right? The problems failed tests report are related to filesystems with 1K block size, so unless I use such filesystems, I'm safe to skip tests and use this new version, especially since I've been using 1.42.7 till now, right?

Or is there some git commit fixing those issues I should try?

Thanks a lot for Your time and sorry about my lame questions..

with best regards

nikola ciprich

Quoted 60 lines · Eric Sandeen

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 14 lines · Eric Sandeen
Quoted 12 lines · Eric Sandeen
Quoted 11 lines · Nikola Ciprich

Hi,

trying to compile latest e2fsprogs and running check under RHEL6-compatible distro (centos) fails:

r_1024_small_bg: ext2 1024 blocksize with small block groups: failed
r_ext4_small_bg: ext4 1024 blocksize with small block groups: failed

dunno whether this is known issue... will bisect help?

with regards

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

nik

Ted, these are the same ones I saw, plus one I think (working on getting all the info).

Sorry this took a while. Attached is a qcow image of a broken r_1024_small_bg filesystem. Doing resize2fs -M on it twice should corrupt it, even on x86_64.

Sorry - the image as attached is not broken, but 2 current resize2fs -M's break it.

- -Eric

Quoted 8 lines

(aside: the test is a bit weird, it does:

echo $RESIZE2FS $RESIZE2FS_OPTS -d 31 -M $TMPFILE $SIZE_2 >> $LOG 2>&1

but specifying -M as well as a size doesn't make much sense?)

-Eric

-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
Comment: GPGTools - http://gpgtools.org
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQIcBAEBAgAGBQJR+HIsAAoJECCuFpLhPd7gT3cP/3YLhpclAhU6UHIW4cqyKhIl STIwV3ROZmic5qQb8X6l5etc8ALCs2b5LVYjhlMlZWISkN64pXOhL0lz7coSABM9 wnZU2y3LntcyGFEg7aEs+sXe2mXqX139NkoNwKfBBPidQ/HdwpnnTdlu4FZeTyrj aDHIUGPs2YBbp2QaBhmzl5b7xwEoHCq/XtrH7v7+SHyHEGzIYTn3w96vyXqVYTjs csUgeMhFFw58Hq+/03UE13Ig/Et8AqSgonnaEiG3ZKXsDIxhjFxqQPGsdD2BleGd Rb/fxnkM0RKhr7xSK3CFWnLQ8LkaUtexazxvr+7BmmgRx8jyQhQj8ASYXzmUuh/S RtbnZ3wioxEwCRhr3FP0ZpoOZwbkDWd/UBy407ZBblxVafjLSYbsMHDM7rPrQHoz oSo1W4c7sy85pNfCgMnQYyuqik5KVDuQNk/nqEkv6tE1YOb5zXTrpDKjEFgBKYVd FpLRk9f0rhoxcwyfZsIY6Wx/4h4rWKnhK6jqL5unDLJitPv405KlNfY28AoFNme4 ZxwDSddausNm+8u9sZON7cuyhAb70Q8N0W6ECPX8+ie8fvXwPDRSJYPpMXqOBK06 35SPqxKlCpT7BGMAqr7+h8qemYdcTS0z7ag2im+MRS1t58AFSsuIyj+IirPWwTUM SQRtuG3mOsWE7l93c8bc
=96jm
-----END PGP SIGNATURE-----

Signature · 13 lines
-------------------------------------
Ing. Nikola CIPRICH
LinuxBox.cz, s.r.o.
28.rijna 168, 709 00 Ostrava

tel.:   +420 591 166 214
fax:    +420 596 621 273
mobil:  +420 777 093 799
www.linuxbox.cz

mobil servis: +420 737 238 656
email servis: [email protected]
-------------------------------------
#9 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 11 lines · Nikola Ciprich

Hello Eric,

I'm very sorry for such a late reply, I was a bit busy with other projects.. Well, I don't really understand what I should do with the images now :) But if I understand previous discussion well, my problem is not caused by code regressions, but new tests which were added to this new version right? The problems failed tests report are related to filesystems with 1K block size, so unless I use such filesystems, I'm safe to skip tests and use this new version, especially since I've been using 1.42.7 till now, right?

Quoted 3 lines

Or is there some git commit fixing those issues I should try?

Thanks a lot for Your time and sorry about my lame questions..

Nah, not lame. To be honest, I've lost track of which resize2fs bugs are fixed and which are not; some have been sent to the list, some have not been fixed, and the e2fsprogs git tree hasn't been updated for over 2 months. So I'm really not sure where things stand right now. :(

In any case, I don't _think_ that this particular bug is yet fixed.

- -Eric

Quoted 27 lines

with best regards

nikola ciprich

On Tue, Jul 30, 2013 at 09:10:53PM -0500, Eric Sandeen wrote:

Quoted 20 lines · Eric Sandeen
Quoted 20 lines
Quoted 20 lines · Eric Sandeen
Quoted 18 lines · Nikola Ciprich
Quoted 5 lines

Hi,

trying to compile latest e2fsprogs and running check under RHEL6-compatible distro (centos) fails:

Quoted 8 lines

r_1024_small_bg: ext2 1024 blocksize with small block groups: failed
r_ext4_small_bg: ext4 1024 blocksize with small block groups: failed

dunno whether this is known issue... will bisect help?

with regards

nik

Ted, these are the same ones I saw, plus one I think (working on getting all the info).

Quoted 18 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Sorry this took a while. Attached is a qcow image of a broken r_1024_small_bg filesystem. Doing resize2fs -M on it twice should corrupt it, even on x86_64.

Sorry - the image as attached is not broken, but 2 current resize2fs -M's break it.

-Eric

Quoted 10 lines
Quoted 8 lines
Quoted 8 lines

(aside: the test is a bit weird, it does:

echo $RESIZE2FS $RESIZE2FS_OPTS -d 31 -M $TMPFILE $SIZE_2 >> $LOG 2>&1

but specifying -M as well as a size doesn't make much sense?)

-Eric

-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
Comment: GPGTools - http://gpgtools.org
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQIcBAEBAgAGBQJSJz1fAAoJECCuFpLhPd7gwxoP/A+0ARl3XX6JJV7SfpVhvFKg qevX6hQnmFD6XgmCnRI1P4SR0NGnfeot8ZIEOGr9W954dENx+9gcF6q6oLUnsrO8 jqBXnUjPIsjYkAD3d5RpK3ygRuOQMMmRtruoSYlEfivG8fYfl0/pbLKDrOhHy/3F 57lsyDwquJhLbIplLu+upe7vctHL9o0QDU0dBVdfrSDTmqD/KOP93qPlK/HfCKip XzTWENnJmLL8WAG6AfbOfWTFFMnKk2qHdglB4VOdNBqiuCLRytScQc5ordwz/Aip NIt9FMcjXQoNgu/IlM8m4gH0UTBkMQDw85cojqbIkMmhUX9gtgUepzMmEKnTViCj iMBM/596dI83Icnh/1nEC/iZ7s0x4yZYwn7l6VTMyW8/gGwYTkGuo4IpD5neeEN6 6b8xRY7YtNdoFlfAWjfssQ7k4JdIPVGX4gJoEyulG4qEJZ70X0R9E8C2tfU3IXSE YsD9fH+w1o0iaNRpuNXpQ1Jy1uA42UHubYgKnAPsHmbAhHS60woC2/+K1lOi4Ree ye1tNAVC9edicNi2L+wLrPwxYnMMv2+x8gs5nfM/PoOPhY8BvqJil2r5QHnsa1q1 cOWQfA9GQ0t+IG9TEgIAQqzdh3QZIs6KEIvDHB1IQk544jVbUG35DOaAPxZsNHJa Cic9x4MFhKsqb+Qb5jw/
=AQPh
-----END PGP SIGNATURE-----