|
From: Sanders, R. M. <san...@za...> - 2006-08-16 12:31:15
|
Nick, Thanks for the reply. I'll apply your diff and see what changes. I did modify my code (as well as the subject line [ forgot yesterday :/ ]) to track all calloc/malloc/realloc/free calls to vet the arguments, but I=20 did *not* see any problems. Stupid other question. I downloaded the valkyrie tarball at the same time as valgrind, and it kernel panic-ed 2 of my dl380's. I don't have the dump messages, but all of my dl380's are running RHEL4 rel 3. Any ideas? Rob -----Original Message----- From: Nicholas Nethercote [mailto:nj...@cs...]=20 Sent: Tuesday, August 15, 2006 6:54 PM To: Sanders, Rob M. Cc: val...@li... Subject: Re: [Valgrind-users] (no subject) On Tue, 15 Aug 2006, Sanders, Rob M. wrote: > A quick question. I just downloaded valgrind 3.2 and build in on our > DL380G2's running RH. I'm seeing a message about 'silly args (-<big=20 > number>) to malloc', but there is no location specified for > *who* is sending the bogus values to calloc. Is this message internal > to valgrind, or a real indication that my code is doing something it=20 > ought not. I'm putting in some checks to my code to verify the args=20 > to malloc/calloc/realloc now to vet my code. It's a real indication. Memcheck doesn't flag it as a "proper" error (with stack trace) because it's very rare and we've been to lazy to make it a proper error -- that requires adding a new error type, making it suppressible, etc. The attached diff will make Memcheck print a stack trace when this warning is shown. Perhaps we should always do that. Or we could do that but only print the warning three times, then say "I won't print it any more". Nick |
|
From: Sanders, R. M. <san...@za...> - 2006-08-16 14:05:07
|
Nick, Many thanks. That patch did indeed point to my problem. Might be = worth=20 including in the baseline. I did tweak it to do essentially the same = thing to=20 the 'complain_about_silly_args2' routine as you did to = 'complain_about_silly_args'. =20 Thanks again.... Rob |
|
From: Sanders, R. M. <san...@za...> - 2006-08-16 20:15:49
|
I'll try to put that info together tomorrow and post it back. I had =
built
everything as myself, but needed to be root to do the make install bits. =
I
want to push the ownership of valkyrie back to a regular user and see if =
that
makes a difference. Other than that, I'm running an almost stock =
RHEL4rel3=20
release on DL380G2. The only update is to the ypbind or ypserv (forget =
which)
required to fix some automount issues. The two machines which exploded =
were NIS clients,
and the valkyrie/valgrind where on a shared common partition. I'll get =
more info
out tomorrow.
Rob
-----Original Message-----
From: val...@li...
[mailto:val...@li...]On Behalf Of Julian
Seward
Sent: Wednesday, August 16, 2006 2:51 PM
To: val...@li...
Cc: Cerion Armour-Brown
Subject: Re: [Valgrind-users] ('silly args')
On Wednesday 16 August 2006 19:31, Cerion Armour-Brown wrote:
> On Wednesday 16 August 2006 14:31, Sanders, Rob M. wrote:
> > Stupid other question. I downloaded the
> > valkyrie
> > tarball at the same time as valgrind, and it kernel panic-ed 2 of my
> > dl380's.
>
> Ouch!
>
> > I don't have the dump messages, but all of my dl380's are running =
RHEL4
> > rel 3.
> > Any ideas?
Valkyrie is a straight C++ Qt application which doesn't do anything
weird (fork/exec/pipe, perhaps) and certainly nothing which a KDE or
Gnome session won't also do. So I think it's pretty strange that this
causes a kernel panic; if that's so then it sounds like a kernel bug.
As Ceri says, do you have any further details?
J
-------------------------------------------------------------------------=
Using Tomcat but need to do more? Need to support web services, =
security?
Get stuff done quickly with pre-integrated technology to make your job =
easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache =
Geronimo
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D=
121642
_______________________________________________
Valgrind-users mailing list
Val...@li...
https://lists.sourceforge.net/lists/listinfo/valgrind-users
|
|
From: Sanders, R. M. <san...@za...> - 2006-08-17 14:32:55
Attachments:
rms001_dir.tgz
|
Ok, as promised, here is a tarball with some of the info requested... It is repeatable - 3 times today even with valkyrie being owned my = myself instead of root I don't have gdb output, but I do have the blurb dumped to = /var/log/messages=20 This indicates a kernel bug. The tarball has the relevent portion of /var/log/message, the output of = 'rpm -qa', a copy of /etc/redhat-release, basic build info for valgrind and valkyrie, and = a little transcript of a session which resulted in the kernel panic contained in message. = Hope this helps... This box is one of 10 NIS client boxes, with the /home partition being = auto-mounted. User=20 accounts and the common directories are contained on this /home = partition. Rob |
|
From: Cerion Armour-B. <ce...@op...> - 2006-08-16 18:31:20
|
On Wednesday 16 August 2006 14:31, Sanders, Rob M. wrote: > Stupid other question. I downloaded the > valkyrie > tarball at the same time as valgrind, and it kernel panic-ed 2 of my > dl380's. Ouch! > I don't have the dump messages, but all of my dl380's are running RHEL4 > rel 3. > Any ideas? 'fraid not - you're the first to report this. can you give more details pls? is it repeatable? where did/does it happen? gdb output / dump message? Cheers, Cerion |
|
From: Julian S. <js...@ac...> - 2006-08-16 18:51:44
|
On Wednesday 16 August 2006 19:31, Cerion Armour-Brown wrote: > On Wednesday 16 August 2006 14:31, Sanders, Rob M. wrote: > > Stupid other question. I downloaded the > > valkyrie > > tarball at the same time as valgrind, and it kernel panic-ed 2 of my > > dl380's. > > Ouch! > > > I don't have the dump messages, but all of my dl380's are running RHEL4 > > rel 3. > > Any ideas? Valkyrie is a straight C++ Qt application which doesn't do anything weird (fork/exec/pipe, perhaps) and certainly nothing which a KDE or Gnome session won't also do. So I think it's pretty strange that this causes a kernel panic; if that's so then it sounds like a kernel bug. As Ceri says, do you have any further details? J |