Re: stress test on ppc

[Thomas Graichen <news-innominate.list.sgi.xfs@xxxxxxxxxxxxx>]

Russell Cattelan <cattelan@xxxxxxxxxxx> wrote:

[ok and now the alpha]

> Finally where in XFS are we trashing AG blocks.

> Ok lets do things in this order
> mkfs a file system
> od -c -N 8 -j 512 /dev/sdb1
> should result in
> 0001000   X   A   G   F  \0  \0  \0 001
> od -c -N 8 -j 1024 /dev/sdb1
> and
>  od -c -N 8 -j 1024 /dev/sdb1
> 0002000   X   A   G   I  \0  \0  \0 001

looks like this last one is wrong? - it's the same as the second

> Those are the magic #s

root@cyan:~# od -c -N 8 -j 512 /dev/sdb1
0001000   X   A   G   F  \0  \0  \0 001
0001010
root@cyan:~# od -c -N 8 -j 1024 /dev/sdb1
0002000   X   A   G   I  \0  \0  \0 001
0002010
root@cyan:~#

> now try
> xfs_db
> sb 0
> print
> agf 0
> print
> agi 0
> print

> save output

root@cyan:~# xfs_db /dev/sdb1
xfs_db: sb 0
xfs_db: print
magicnum = 0x58465342
blocksize = 8192
dblocks = 33130
rblocks = 0
rextents = 0
uuid = eac0a48b-4dff-41c3-b2cc-e110bd5c3ff8
logstart = 32772
rootino = 256
rbmino = 257
rsumino = 258
rextsize = 8
agblocks = 4142
agcount = 8
rbmblocks = 0
logblocks = 1000
versionnum = 0x2084
sectsize = 512
inodesize = 256
inopblock = 32
fname = "\000\000\000\000\000\000\000\000\000\000\000\000"
blocklog = 13
sectlog = 9
inodelog = 8
inopblog = 5
agblklog = 13
rextslog = 0
inprogress = 0
imax_pct = 25
icount = 64
ifree = 61
fdblocks = 32096
frextents = 0
uquotino = 0
pquotino = 0
qflags = 0
flags = 0
shared_vn = 0
inoalignmt = 1
unit = 0
width = 0
dirblklog = 0
xfs_db: agf 0
xfs_db: print
magicnum = 0x58414746
versionnum = 1
seqno = 0
length = 4142
bnoroot = 1
cntroot = 2
bnolevel = 1
cntlevel = 1
flfirst = 0
fllast = 3
flcount = 4
freeblks = 4132
longest = 4132
xfs_db: agi 0
xfs_db: print
magicnum = 0x58414749
versionnum = 1
seqno = 0
length = 4142
count = 64
root = 3
level = 1
freecount = 61
newino = 256
dirino = null
unlinked[0-63] = 
xfs_db: quit
root@cyan:~#

> Mount the file system
> run od commands and xfs_db command again
> save output

root@cyan:~# od -c -N 8 -j 512 /dev/sdb1
0001000   X   A   G   F  \0  \0  \0 001
0001010
root@cyan:~# od -c -N 8 -j 1024 /dev/sdb1
0002000   X   A   G   I  \0  \0  \0 001
0002010
root@cyan:~# xfs_db -r /dev/sdb1
xfs_db: sb 0
xfs_db: print
magicnum = 0x58465342
blocksize = 8192
dblocks = 33130
rblocks = 0
rextents = 0
uuid = eac0a48b-4dff-41c3-b2cc-e110bd5c3ff8
logstart = 32772
rootino = 256
rbmino = 257
rsumino = 258
rextsize = 8
agblocks = 4142
agcount = 8
rbmblocks = 0
logblocks = 1000
versionnum = 0x2084
sectsize = 512
inodesize = 256
inopblock = 32
fname = "\000\000\000\000\000\000\000\000\000\000\000\000"
blocklog = 13
sectlog = 9
inodelog = 8
inopblog = 5
agblklog = 13
rextslog = 0
inprogress = 0
imax_pct = 25
icount = 64
ifree = 61
fdblocks = 32096
frextents = 0
uquotino = 0
pquotino = 0
qflags = 0
flags = 0
shared_vn = 0
inoalignmt = 1
unit = 0
width = 0
dirblklog = 0
xfs_db: agf 0
xfs_db: print
magicnum = 0x58414746
versionnum = 1
seqno = 0
length = 4142
bnoroot = 1
cntroot = 2
bnolevel = 1
cntlevel = 1
flfirst = 0
fllast = 3
flcount = 4
freeblks = 4132
longest = 4132
xfs_db: agi 0
xfs_db: print
magicnum = 0x58414749
versionnum = 1
seqno = 0
length = 4142
count = 64
root = 3
level = 1
freecount = 61
newino = 256
dirino = null
unlinked[0-63] = 
xfs_db:

> Now do some FS activity
> run tod command and xfs_sb commands again
> save output.

root@cyan:~# od -c -N 8 -j 512 /dev/sdb1
0001000  \0  \0  \0  \0 377 377 377 377
0001010
root@cyan:~# od -c -N 8 -j 1024 /dev/sdb1
0002000   X   A   G   I  \0  \0  \0 001
0002010
root@cyan:~# xfs_db -r /dev/sdb1
xfs_db: sb 0
xfs_db: print
magicnum = 0x58465342
blocksize = 8192
dblocks = 33130
rblocks = 0
rextents = 0
uuid = 89cb7e3c-c0cc-4a42-8013-aef1f692b80b
logstart = 32772
rootino = 256
rbmino = 257
rsumino = 258
rextsize = 8
agblocks = 4142
agcount = 8
rbmblocks = 0
logblocks = 1000
versionnum = 0x2084
sectsize = 512
inodesize = 256
inopblock = 32
fname = "\000\000\000\000\000\000\000\000\000\000\000\000"
blocklog = 13
sectlog = 9
inodelog = 8
inopblog = 5
agblklog = 13
rextslog = 0
inprogress = 0
imax_pct = 25
icount = 2240
ifree = 202
fdblocks = 8423
frextents = 0
uquotino = 0
pquotino = 0
qflags = 0
flags = 0
shared_vn = 0
inoalignmt = 1
unit = 0
width = 0
dirblklog = 0
xfs_db: agf 0
xfs_db: print
magicnum = 0
versionnum = 4294967295
seqno = 0
length = 0
bnoroot = 2966461184
cntroot = 16580607
bnolevel = 16580607
cntlevel = 0
flfirst = 2147483648
fllast = 16777216
flcount = 3223092738
freeblks = 16580607
longest = 32258
xfs_db:agi 0
xfs_db: print
magicnum = 0x58414749
versionnum = 1
seqno = 0
length = 4142
count = 448
root = 3
level = 1
freecount = 0
newino = 122016
dirino = null
unlinked[0-63] = 
xfs_db:

> Send all the output to us.

i hope it is ok to post it here - i think it's not too much - but if
anyone does not like those big debugging mails i can also upload
them to a ftp accessable place in the future

> This should tell us where we need to start fixing things first.

> I suspect we will need to find an alpha around someplace to actually
> get some of this stuff debugged.

i'll update the alpha to the latest kernel now and recheck that it
is still there (you never know :-) ... also keep in mind: compiler
bugs are possible too ... it's gcc 2.95.2 which has problems on
intel too but seems to work fine on the ppc - but this kind
of problem does not really look like a compiler bug to me
on the other side

good luck

t

-- 
thomas.graichen@xxxxxxxxxxxxxx
technical director                                       innominate AG
clustering & security                             the linux architects
tel: +49-30-308806-13   fax: -77             http://www.innominate.com