anaconda segfault in rawhide 20050917 x86_64 nfs install

[Willem Riede]

I guess there's progress, but not much :-)

With Saturday's (17th) rawhide on my opteron I get a segfault a couple of
seconds into anaconda using nfs mount of the x86_64 rawhide tree.

<Alt-F3> last three lines are:

Display mode = g
Method = nfs://mnt/source/.
_Fedora is only installclass, using it

<Alt-F4> reveals:

audit(1127046737.776:2): avc: denied { associate } for pid=733    \
  comm="anaconda" name="rpm" scontext=system_u:object_r:root_t:s0 \
  tcontext=system_u:object_r:root_t:s0 tsclass=filesystem
loader[733]: segfault at 0000000000000000 rip 00002aaaab3019f0    \
  rsp 00007ffff8f7d58 error 4

Does anybody else see this?

Regards, Willem Riede.

[Jeremy Katz]

On Sun, 2005-09-18 at 12:52 +0000, Willem Riede wrote:
>  I guess there's progress, but not much :-) 
There's lots of things changing :)   Unfortunately, testing the full
tree effect is pretty time consuming, so we've been doing a lot of spot
testing as we add stuff and then wait until the next day for the "whole
tree" effect.  Unfortunately, given the amount of change, this is
somewhat prone to causing problems :/

>  With Saturday's (17th) rawhide on my opteron I get a segfault a
> couple of
>  seconds into anaconda using nfs mount of the x86_64 rawhide tree. 
Hmm, that means that anaconda is running and it looks like we're
probably into the X probing.  Could be a side effect of the kudzu
changes to return NULL in some cases.  I'm not sure if I'll get a chance
to prod at it today, but I'll try.

Jeremy

[Willem Riede]

On 09/18/2005 11:04:35 AM, Jeremy Katz wrote:
>  On Sun, 2005-09-18 at 12:52 +0000, Willem Riede wrote:
>  > I guess there's progress, but not much :-)
>  
>  There's lots of things changing :)   Unfortunately, testing the full
>  tree effect is pretty time consuming, so we've been doing a lot of spot
>  testing as we add stuff and then wait until the next day for the "whole
>  tree" effect.  Unfortunately, given the amount of change, this is
>  somewhat prone to causing problems :/ 
I do understand that - progress will come in many small steps.
If there is anything I can do to help (test), let me know, in
the mean time I'll continue providing you data points.

>  > With Saturday's (17th) rawhide on my opteron I get a
> segfault a couple of
>  > seconds into anaconda using nfs mount of the x86_64 rawhide tree.
>  
>  Hmm, that means that anaconda is running and it looks like we're
>  probably into the X probing.  Could be a side effect of the kudzu
>  changes to return NULL in some cases.  I'm not sure if I'll get a chance
>  to prod at it today, but I'll try. 
FWIW, today's rawhide has the same problem, but that version was probably
already in the pipeline before this mail exchange.

Thanks, Willem Riede.

[Jeremy Katz]

On Sun, 2005-09-18 at 21:24 +0000, Willem Riede wrote:
>  On 09/18/2005 11:04:35 AM, Jeremy Katz wrote:
>  > On Sun, 2005-09-18 at 12:52 +0000, Willem Riede wrote:
>  > > I guess there's progress, but not much :-)
>  > 
>  > There's lots of things changing :)   Unfortunately, testing the full
>  > tree effect is pretty time consuming, so we've been doing a lot of spot
>  > testing as we add stuff and then wait until the next day for the "whole
>  > tree" effect.  Unfortunately, given the amount of change, this is
>  > somewhat prone to causing problems :/
>  
>  I do understand that - progress will come in many small steps.
>  If there is anything I can do to help (test), let me know, in
>  the mean time I'll continue providing you data points. 
*nod*  Hopefully we'll be ready for some more targeted testing soon...
there's a light at the end of this tunnel, and I don't think it's an
oncoming train ;)

Jeremy