Evolution and glibc on x86_64

Ola Thoresen redhat at olen.net
Thu Aug 24 12:58:23 UTC 2006


Quoting PFJ <paul at all-the-johnsons.co.uk>:

> Hi,
>
> I'm running rawhide on an x86_64 box. After applying yesterdays updates
> for Evolution and glibc, I'm finding Evolution has stopped working
> completely.
>
> I start evolution up via either the command line or by selecting the
> icon, everything is okay and then things go wrong. Evolution starts
> eating upto 90% of the processor time and refuses to go any further.
>
>> From the command line, I've attempted to start it up with and without
> plugins. The software hasn't crashed, it's just got itself confused.
>
> Is anyone else seeing this?

Yes.
Same problem here.

I have done some debugging, and there seems to be two (unrelated?) problems.
1) Starting evolution hangs in an endless loop of

mmap(NULL, 68719480832, PROT_READ|PROT_WRITE,  
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = -1
ENOMEM (Cannot allocate memory)
mmap(NULL, 68719480832, PROT_READ|PROT_WRITE,  
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = -1
ENOMEM (Cannot allocate memory)
mmap(NULL, 68719480832, PROT_READ|PROT_WRITE,  
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = -1
ENOMEM (Cannot allocate memory)

This comes just after it tries to read any  
~/.evolution/mail/local/*.cmeta-file

Deleting all the .cmeta-files will at least let evolution start.


2) After starting up evolution (deleting the .cmeta-files) it is still  
unusable.
It is using 90% cpu, but now there is no apparent reason for the hang.
After abount 30 minutes of waiting for any folder to show up, I just  
had to kill it to be
able to use the machine at all.

An strace of the process now onlys shows

poll([{fd=5, events=POLLIN}, {fd=4, events=POLLIN, revents=POLLIN}, {fd=9,
events=POLLIN|POLLPRI}, {fd=11, events=POLLIN|POLLPRI}, {fd=27,  
events=POLLIN}, {fd=19,
events=POLLIN}, {fd=23, events=POLLIN}], 7, 12952) = 1
ioctl(4, FIONREAD, [32])                = 0
read(4, "\241 U7z\0\300\2\17\1\0\0\26\1\0\0\254\r\0\0z\0\300\2\0"..., 32) = 32
write(4, "\31\0\v\0M\0\0\0\0\0\30\0! f\0M\0\0\0\17\1\0\0\26\1\0\0"...,  
44) = 44
ioctl(4, FIONREAD, [0])                 = 0
ioctl(4, FIONREAD, [0])                 = 0
poll([{fd=5, events=POLLIN}, {fd=4, events=POLLIN, revents=POLLIN}, {fd=9,
events=POLLIN|POLLPRI}, {fd=11, events=POLLIN|POLLPRI}, {fd=27,  
events=POLLIN}, {fd=19,
events=POLLIN}, {fd=23, events=POLLIN}], 7, 7924) = 1
ioctl(4, FIONREAD, [32])                = 0
read(4, "\241 V7z\0\300\2\17\1\0\0\26\1\0\0\255\r\0\0z\0\300\2\0"..., 32) = 32
write(4, "\31\0\v\0M\0\0\0\0\0\30\0! f\0M\0\0\0\17\1\0\0\26\1\0\0"...,  
44) = 44
ioctl(4, FIONREAD, [0])                 = 0
ioctl(4, FIONREAD, [0])                 = 0

With a new poll about every 2 or 3 seconds.

Any bugzilla entry I should add this to?

Rgds.

Ola Thoresen







More information about the fedora-test-list mailing list