Created attachment 1098635 [details]
resize2fs: fix off-line resize of file systems with flex_bg && !resize_inode

Description of problem:
When trying to resize the new partition from very small size to a size bigger than the origin 1000 times, i got the problem when using the command fsck.

For example: creating a new logical volume of 16M, after that, resizing it to 20G. 

I found that it came from the command "resize2fs"

Version of e2fsprogs that cause the problem : 1.41.12 -> 1.42.7 

Version OS: RHEL 6 (6.7) or Centos 6 (6.7)


How reproducible:


Steps to Reproduce:
1.  export VG=vg_centos6
2.  lvcreate -L16M -n test $VG
3.  mkfs.ext4 /dev/$VG/test
4.  fsck -fn /dev/$VG/test   // no problem 
5.  lvresize -L20G /dev/$VG/test
6.  resize2fs /dev/$VG/test  
7.  fsck -fn /dev/$VG/test  // check again and this command shows a lot of errors

Actual results:

2fsck 1.41.12 (17-May-2010)
fsck.ext4: Group descriptors look bad... trying backup blocks...
Pass 1: Checking inodes, blocks, and sizes
Group 2's block bitmap at 84 conflicts with some other fs block.
Relocate? no

Group 14's block bitmap at 81 conflicts with some other fs block.
Relocate? no

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Block bitmap differences:  -(16385--20621) -114689 -(114707--114710) -114713 -114721 -114725 -(114739--114742) -114745 -114753 -114758 -(114771--114774) -114777 -(114790--114796) -114812 -114833 -114835 -114929 -114931 -114934 -114936 -(114943--114944) -114946 -(114963--114966) -114969 -114978 -114981 -(114995--114998) -115001 -115009 -115014 -115017 -(115027--115030) -115033 -115054 -115068 -(115089--115091) -(115186--115189) -115191 -115195 -115199 -(115201--115202) -(115219--115222) -115225 -(115233--115234) -115237 -(115251--115254) -115257 -115265 -115270 -115274 -(115283--115286) -115289 -115310 -115324 -(115345--115347) -115443 -115445 -(115448--115452) -115454 -115456 -115459 -(115475--115478) -115481 -115491 -115493 -(115507--115510) -115513 -115521 -11552 
....
(15869953--15876096) -(16001025--16007168) -(16132097--16138240) -(16263169--16269312) -(16394241--16400384) -(16525313--16531456) -(16656385--16662528)
Fix? no

Free blocks count wrong for group #2 (8192, counted=3955).
Fix? no

Free blocks count wrong for group #14 (8192, counted=6932).
Fix? no

Free blocks count wrong for group #177 (8192, counted=2048).
Fix? no

Free blocks count wrong for group #193 (8192, counted=2048).
Fix? no

......
Free blocks count wrong (20308626, counted=19590425).
Fix? no


/dev/mapper/datavg-test: ***** FILE SYSTEM WAS MODIFIED *****

/dev/mapper/datavg-test: ********** WARNING: Filesystem still has errors **********

/dev/mapper/datavg-test: 11/5242880 files (9.1% non-contiguous), 662894/20971520 blocks
Error writing block 2 (Attempt to write block from filesystem resulted in short write).  Ignore error? no

Error writing block 3 (Attempt to write block from filesystem resulted in short write).  Ignore error? no

Error writing block 4 (Attempt to write block from filesystem resulted in short write).  Ignore error? no

Error writing block 5 (Attempt to write block from filesystem resulted in short write).  Ignore error? no

Expected results:

fsck does pass all the check like the following :

e2fsck 1.41.12 (17-May-2010)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/vg_centos6/test: 11/5242880 files (9.1% non-contiguous), 662896/20971520 blocks


Additional info:

i have already created a patch for this bug. Actually, this bug has been fixed in version 1.42.8 of e2fsprogs on centos 7 but it is not backported to centos 6.7.

What i did in the patch is to bring the modification from the commit "f026f1a37061a021f7188c9e45d0fbff1bde9e60" and ajust some places in order to make it work with the current version of e2fsprogs.

You can find my patch in file attachment.