* [uml-devel] uml-patch-2.4.23-2
@ 2004-01-17 15:47 Jeff Dike
2004-01-20 9:20 ` Nick Craig-Wood
0 siblings, 1 reply; 6+ messages in thread
From: Jeff Dike @ 2004-01-17 15:47 UTC (permalink / raw)
To: user-mode-linux-devel
This patch fixes a number of bugs including
with ubd-mmap in force, device data remaining mapped after the device
was closed - this should fix the EBADF errors some people were seeing on UML
shutdown, it allows 2.6 to boot with ubd-mmap, and there's some potential that
it fixed file corruption. Anyone who has been seeing ubd-mmap problems should
test this and let me know one way or the other.
/dev/anon is now correctly checked for and used if it's available
fixed the linker script to get rid of the segfaults seen with RH and
Fedora
mhz is now a long long, so UML again will stop hanging on fast hosts
lots of small changes to fix minor bugs, compilation problems, and
to reduce the number of diffs between the 2.4 and 2.6 trees
CVS hasn't been updated yet because SF is broken again. I'll check everything
in when the CVS server is back up.
Jeff
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [uml-devel] uml-patch-2.4.23-2
2004-01-17 15:47 [uml-devel] uml-patch-2.4.23-2 Jeff Dike
@ 2004-01-20 9:20 ` Nick Craig-Wood
2004-01-20 15:00 ` Jeff Dike
0 siblings, 1 reply; 6+ messages in thread
From: Nick Craig-Wood @ 2004-01-20 9:20 UTC (permalink / raw)
To: Jeff Dike; +Cc: user-mode-linux-devel
I looked for this on sourceforge - it is linked to on the projects
page, but it gives a 404 error when you try to download it. I had a
look for it on uploads.sourceforge.net also (in the user-mode-linux
directory and in /incoming) and its not there either!
--
Nick Craig-Wood
ncw1@axis.demon.co.uk
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [uml-devel] uml-patch-2.4.23-2
2004-01-20 9:20 ` Nick Craig-Wood
@ 2004-01-20 15:00 ` Jeff Dike
2004-01-20 17:35 ` Nick Craig-Wood
0 siblings, 1 reply; 6+ messages in thread
From: Jeff Dike @ 2004-01-20 15:00 UTC (permalink / raw)
To: Nick Craig-Wood; +Cc: user-mode-linux-devel
ncw1@axis.demon.co.uk said:
> I looked for this on sourceforge - it is linked to on the projects
> page, but it gives a 404 error when you try to download it.
Hmmm, I thought I did the SF upload dance, but maybe I forgot. In any case,
it's there now.
> I had a
> look for it on uploads.sourceforge.net also (in the user-mode-linux
> directory and in /incoming) and its not there either!
Things get deleted from there in a few hours or so.
Jeff
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [uml-devel] uml-patch-2.4.23-2
2004-01-20 15:00 ` Jeff Dike
@ 2004-01-20 17:35 ` Nick Craig-Wood
2004-01-20 20:40 ` Jeff Dike
0 siblings, 1 reply; 6+ messages in thread
From: Nick Craig-Wood @ 2004-01-20 17:35 UTC (permalink / raw)
To: Jeff Dike; +Cc: user-mode-linux-devel
On Tue, Jan 20, 2004 at 10:00:09AM -0500, Jeff Dike wrote:
> Hmmm, I thought I did the SF upload dance, but maybe I forgot. In any case,
> it's there now.
I have it now thanks!
I've discovered an easy way to re-create the data corruption with
ubd=mmap (with or without /dev/anon) using the fsx tool from here.
http://www.freebsd.org/cgi/cvsweb.cgi/src/tools/regression/fsx/
Download fxs.c, compile it
gcc -Wall -O3 -o fsx fsx.c
and run it
./fsx z
With ubd=mmap it blows up after less than a minute usually. Without
it seems to run fine.
It doesn't always blow up in the same place (fsx is by default
deterministic) indicating that this is a racey sort of problem...
I diffed the hex dump to see what sort of corruptions they were.
Nothing obvious springs to mind, other than that the corruptions start
on a page boundary, but don't extend for a whole page. Maybe the page
is being re-used before writeout has finished?
--- z.fsxgood.hex 2004-01-20 17:25:20.000000000 +0000
+++ z.hex 2004-01-20 17:25:13.000000000 +0000
@@ -7422,26 +7422,26 @@
001cfd0: cc fa cc e1 cc 16 cc 6d cc 36 cc 6f cc de cc 32 .......m.6.o...2
001cfe0: cc 33 cc bc cc 7f cc 14 cc 1e cc 19 cc 78 cc 2a .3...........x.*
001cff0: cc 55 cc 2d cc ec cc ee cc e1 cc 6e cc 12 cc 68 .U.-.......n...h
-001d000: cc 80 cc 12 cc 54 cc c8 cc 4f cc 0e cc 1e cc d1 .....T...O......
-001d010: cc 2a cc 43 cc 7d cc 1e cc 06 cc a5 cc 42 cc cb .*.C.}.......B..
-001d020: cc 09 cc bc cc b1 cc 06 cc 65 cc 94 cc f6 cc b8 .........e......
-001d030: cc 64 cc 00 cc 41 cc 46 cc e7 cc 1f cc c1 cc 21 .d...A.F.......!
-001d040: cc c4 cc 40 cc 72 cc 26 cc 62 cc 8e cc 7c cc b5 ...@.r.&.b...|..
-001d050: cc 86 cc c0 cc c4 cc d2 cc ea cc af cc 01 cc 67 ...............g
-001d060: cc d2 cc 67 cc 79 cc 3f cc c0 cc fc cc ae cc 0f ...g.y.?........
-001d070: cc a1 cc 2a cc 36 cc 40 cc 18 cc 05 cc 8c cc ea ...*.6.@........
-001d080: cc 21 cc 93 cc dc cc 14 cc 82 cc 58 cc bf cc f0 .!.........X....
-001d090: cc dc cc 76 cc b9 cc 00 cc 17 cc 54 cc be cc 31 ...v.......T...1
-001d0a0: cc 2f cc 4a cc de cc 5e cc 0e cc 98 cc 49 cc 69 ./.J...^.....I.i
-001d0b0: cc 48 cc 79 cc c8 cc cc cc b8 cc 20 cc f6 cc 2b .H.y....... ...+
-001d0c0: cc 4b cc 7f cc fb cc 22 cc 51 cc 1a cc ea cc ec .K.....".Q......
-001d0d0: cc 01 cc 4e cc 2e cc b6 cc 2f cc 75 cc f4 cc c5 ...N...../.u....
-001d0e0: cc 26 cc cf cc 67 cc 80 cc 35 cc b0 cc ef cc fe .&...g...5......
-001d0f0: cc 54 cc 19 cc 86 cc d4 cc e6 cc 4e cc c1 cc d5 .T.........N....
-001d100: cc 9a cc 12 cc b8 cc 1b cc b1 cc 41 cc a1 cc f3 ...........A....
-001d110: cc 08 cc f9 cc f2 cc 11 cc 60 cc a4 cc 5a cc b0 .........`...Z..
-001d120: cc e3 cc fc cc c4 cc 6f cc 0d cc 54 cc 37 cc 18 .......o...T.7..
-001d130: cc d6 cc 6c cc e2 cc de c8 62 c8 91 c8 db c8 6d ...l.....b.....m
+001d000: c8 7c c8 0e c8 50 c8 c4 c8 4b c8 0a c8 1a c8 cd .|...P...K......
+001d010: c8 26 c8 3f c8 79 c8 1a c8 02 c8 a1 c8 3e c8 c7 .&.?.y.......>..
+001d020: c8 05 c8 b8 c8 ad c8 02 c8 61 c8 90 c8 f2 c8 b4 .........a......
+001d030: c8 60 c8 fc c8 3d c8 42 c8 e3 c8 1b c8 bd c8 1d .`...=.B........
+001d040: c8 c0 c8 3c c8 6e c8 22 c8 5e c8 8a c8 78 c8 b1 ...<.n.".^...x..
+001d050: c8 82 c8 bc c8 c0 c8 ce c8 e6 c8 ab c8 fd c8 63 ...............c
+001d060: c8 ce c8 63 c8 75 c8 3b c8 bc c8 f8 c8 aa c8 0b ...c.u.;........
+001d070: c8 9d c8 26 c8 32 c8 3c c8 14 c8 01 c8 88 c8 e6 ...&.2.<........
+001d080: c8 1d c8 8f c8 d8 c8 10 c8 7e c8 54 c8 bb c8 ec .........~.T....
+001d090: c8 d8 c8 72 c8 b5 c8 fc c8 13 c8 50 c8 ba c8 2d ...r.......P...-
+001d0a0: c8 2b c8 46 c8 da c8 5a c8 0a c8 94 c8 45 c8 65 .+.F...Z.....E.e
+001d0b0: c8 44 c8 75 c8 c4 c8 c8 c8 b4 c8 1c c8 f2 c8 27 .D.u...........'
+001d0c0: c8 47 c8 7b c8 f7 c8 1e c8 4d c8 16 c8 e6 c8 e8 .G.{.....M......
+001d0d0: c8 fd c8 4a c8 2a c8 b2 c8 2b c8 71 c8 f0 c8 c1 ...J.*...+.q....
+001d0e0: c8 22 c8 cb c8 63 c8 7c c8 31 c8 ac c8 eb c8 fa ."...c.|.1......
+001d0f0: c8 50 c8 15 c8 82 c8 d0 c8 e2 c8 4a c8 bd c8 d1 .P.........J....
+001d100: c8 96 c8 0e c8 b4 c8 17 c8 ad c8 3d c8 9d c8 ef ...........=....
+001d110: c8 04 c8 f5 c8 ee c8 0d c8 5c c8 a0 c8 56 c8 ac .........\...V..
+001d120: c8 df c8 f8 c8 c0 c8 6b c8 09 c8 50 c8 33 c8 14 .......k...P.3..
+001d130: c8 d2 c8 68 c8 de c8 da c8 62 c8 91 c8 db c8 6d ...h.....b.....m
001d140: c8 bd c8 78 c8 a7 c8 ca c8 e4 c8 ed c8 b8 c8 3c ...x...........<
001d150: c8 36 c8 f2 c8 29 c8 18 c8 ea c8 c3 c8 d8 c8 fb .6...)..........
001d160: c8 2c c8 27 c8 cc c8 7b c8 e3 c8 16 c8 1e c8 3f .,.'...{.......?
--
Nick Craig-Wood
ncw1@axis.demon.co.uk
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [uml-devel] uml-patch-2.4.23-2
2004-01-20 17:35 ` Nick Craig-Wood
@ 2004-01-20 20:40 ` Jeff Dike
2004-01-21 7:41 ` Nick Craig-Wood
0 siblings, 1 reply; 6+ messages in thread
From: Jeff Dike @ 2004-01-20 20:40 UTC (permalink / raw)
To: Nick Craig-Wood; +Cc: user-mode-linux-devel
ncw1@axis.demon.co.uk said:
> I diffed the hex dump to see what sort of corruptions they were.
> Nothing obvious springs to mind, other than that the corruptions start
> on a page boundary, but don't extend for a whole page. Maybe the page
> is being re-used before writeout has finished?
Interesting, but I'm not sure what it means. I had a 100% reproducable
test last night, thanks to Chris Aker, but it became somewhat iffy today.
BTW, what filesystem is this? I'm seeing corruption with ext3, but nothing
with ext2.
Jeff
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [uml-devel] uml-patch-2.4.23-2
2004-01-20 20:40 ` Jeff Dike
@ 2004-01-21 7:41 ` Nick Craig-Wood
0 siblings, 0 replies; 6+ messages in thread
From: Nick Craig-Wood @ 2004-01-21 7:41 UTC (permalink / raw)
To: Jeff Dike; +Cc: user-mode-linux-devel
On Tue, Jan 20, 2004 at 03:40:11PM -0500, Jeff Dike wrote:
> ncw1@axis.demon.co.uk said:
> > I diffed the hex dump to see what sort of corruptions they were.
> > Nothing obvious springs to mind, other than that the corruptions start
> > on a page boundary, but don't extend for a whole page. Maybe the page
> > is being re-used before writeout has finished?
>
> Interesting, but I'm not sure what it means.
fxs tells you what operation caused the problem. Its output is a
little hard to decode though!
> I had a 100% reproducable test last night, thanks to Chris Aker, but
> it became somewhat iffy today.
fxs has blown up within a minute each time I've tried it with
ubd=mmap. Not always in the same place though!
FWIW I left fxs running on 2.4.23-2 without ubd=mmap and it ran fine
all night.
> BTW, what filesystem is this? I'm seeing corruption with ext3, but nothing
> with ext2.
Its ext2
/dev/root / ext2 rw,noatime,nodiratime 0 0
--
Nick Craig-Wood
ncw1@axis.demon.co.uk
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-01-21 7:41 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-01-17 15:47 [uml-devel] uml-patch-2.4.23-2 Jeff Dike
2004-01-20 9:20 ` Nick Craig-Wood
2004-01-20 15:00 ` Jeff Dike
2004-01-20 17:35 ` Nick Craig-Wood
2004-01-20 20:40 ` Jeff Dike
2004-01-21 7:41 ` Nick Craig-Wood
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.