Kernel zombie threads after module removal.
This is very close to the problem related in http://lkml.org/archive/2002/2/4/368/index.html
but I have no USB. I have SCSI with aha152x_cs.o,
and after doing "cardctl eject" that removes the module,
the process scsi_eh_0 stays as zombie.
If I repeat the operation of card insert and eject,
I get several scsi_eh_0 zombies.
Now with 2.4.18, but I have got this problem with others 2.4 kernels.
Quoted 5 lines · Christian.Gennerat
This is very close to the problem related in
http://lkml.org/archive/2002/2/4/368/index.html
but I have no USB. I have SCSI with aha152x_cs.o, and after doing "cardctl eject" that removes the module, the processscsi_eh_0stays as zombie.
Add
reparent_to_init();
after the call to daemonize() in scsi_error_handler() in drivers/scsi/scsi_error.c
Disclaimer: I don't know this code at all
regards
john
Signature · 2 lines
"If you look 'round the table and can't tell who the sucker is, it's you." - Quiz Show
PATCH 1/1 Kernel zombie threads after module removal.
In reply to Christian.GenneratQuoted 6 lines · John Levon
Quoted 5 lines
This is very close to the problem related in http://lkml.org/archive/2002/2/4/368/index.html
but I have no USB. I have SCSI with aha152x_cs.o,
and after doing "cardctl eject" that removes the module,
the processscsi_eh_0stays as zombie.
Quoted 9 lines
Add
reparent_to_init();after the call to
daemonize()inscsi_error_handler()indrivers/scsi/scsi_error.cDisclaimer: I don't know this code at all
Good.
scsi_error.c renamed from scsi_error.c.orig +0 -0
@@ -1860,6 +1860,7 @@
*/ daemonize();
+ reparent_to_init();
/* * Set the name of this process.