[Parted-maintainers] Bug#602568: squeeze-di-beta1 installer: partman hang

[Jean-Christian de Rivaz]

Colin Watson a écrit :
> On Tue, Nov 09, 2010 at 07:46:37PM +0100, Jean-Christian de Rivaz wrote:
>> So even if the primary cause is now fixed, from a quality view, this bug  
>> is still not in the state to be closed. Or should other bug be opened  
>> for those point ?
> 
> If you believe parted_server (which is in the partman-base package)
> needs improvements, please clone this bug and reassign the clone to
> partman-base.  Keeping this bug open for that would make tracking too
> difficult.
> 

Ok. I can now reproduce the partman-base problem, using a specially 
created crash prone /lib/libparted.so.0.0.1 library. The parted-server, 
as expected, crash this way on the console it has been started from:

root at point:/home/jcdr/partman-base-test# bin/parted_server
root at point:/home/jcdr/partman-base-test# echo "OPEN =dev=sdb /dev/sdb" > 
/var/lib/partman/infifo
root at point:/home/jcdr/partman-base-test# Backtrace has 19 calls on stack:
   19: /lib/libparted.so.0(ped_assert+0x28) [0xb77bcea3]
   18: /lib/libparted.so.0(ped_geometry_new+0x5a) [0xb77c6909]
   17: /lib/libparted.so.0(ped_geometry_duplicate+0x73) [0xb77c69e2]
   16: /lib/libparted.so.0(ped_constraint_init+0x15e) [0xb77c7aae]
   15: /lib/libparted.so.0(ped_constraint_new+0x82) [0xb77c7b70]
   14: /lib/libparted.so.0(+0x4eef7) [0xb77feef7]
   13: /lib/libparted.so.0(+0x4f60c) [0xb77ff60c]
   12: /lib/libparted.so.0(+0x4fc78) [0xb77ffc78]
   11: /lib/libparted.so.0(+0x501e6) [0xb78001e6]
   10: /lib/libparted.so.0(+0x130ab) [0xb77c30ab]
   9: /lib/libparted.so.0(ped_disk_add_partition+0x15f) [0xb77c5896]
   8: /lib/libparted.so.0(+0x4d1e4) [0xb77fd1e4]
   7: /lib/libparted.so.0(+0x4d406) [0xb77fd406]
   6: /lib/libparted.so.0(ped_disk_new+0xd2) [0xb77c1bda]
   5: bin/parted_server() [0x805594d]
   4: bin/parted_server() [0x8055b4b]
   3: bin/parted_server() [0x805621b]
   2: /lib/i686/cmov/libc.so.6(__libc_start_main+0xe6) [0xb7680c76]
   1: bin/parted_server() [0x804a101]

The /var/lib/partman/outfifo simply report:
OK
OK

And /var/log/partman say:

parted_server: ======= Starting the server
parted_server: main_loop: iteration 1
parted_server: Opening infifo
parted_server: Read command: OPEN
parted_server: command_open()
parted_server: Request to open =dev=sdb
parted_server: Opening outfifo
parted_server: OUT: OK


parted_server: OUT: OK


So I think it's not really parted_server the cause of the problem (is 
case of libparted problem), but more the script that started it and 
failed to observe that it crashed.

Still, this script probably in partman-base too.

But what do you say by "clone this bug" ?
* It there a procedure to do this ?
* Or it's a administrative option I don't have access to as a user ?
* Or I have to manually copy the report into a new one ?

Regards,

Jean-Christian de Rivaz