Re: multipathd segfault

[Hannes Reinecke <hare@suse.de>]

[-- Attachment #1: Type: text/plain, Size: 2351 bytes --]

David Strand wrote:
> I am getting a multipathd segfault when I have more than 26 devices,
> and the device names become 4 characters. (i.e. /dev/sdz is consumed
> and the next device is /dev/sdaa).
> 
> I was trying to hook gdb to a core dump, but it keeps complaining that
> it needs device-mapper-libs-debuginfo, which I can not find. Can
> anyone point me to this?
> 
> I built multipathd with -g turned on and did not strip it. Here is the
> last few lines of output when I run it as multipathd -v9 -d:
> 
> Jul 04 05:01:26 | alpa-0x000080: discover
> Jul 04 05:01:26 | libdevmapper: ioctl/libdm-iface.c(1606): dm table
> alpa-0x000080  NF   [16384]
> Jul 04 05:01:26 | *word = 0, len = 1
> Jul 04 05:01:26 | *word = 0, len = 1
> Jul 04 05:01:26 | *word = 1, len = 1
> Jul 04 05:01:26 | *word = 1, len = 1
> Jul 04 05:01:26 | *word = 1, len = 1
> Jul 04 05:01:26 | *word = 1, len = 1
> Jul 04 05:01:26 | *word = 65:160, len = 6
> Jul 04 05:01:26 | *word = 1000, len = 4
> Jul 04 05:01:26 | libdevmapper: ioctl/libdm-iface.c(1606): dm status
> alpa-0x000080  NF   [16384]
> Jul 04 05:01:26 | *word = 2, len = 1
> Jul 04 05:01:26 | *word = 0, len = 1
> Jul 04 05:01:26 | *word = 0, len = 1
> Jul 04 05:01:26 | *word = 1, len = 1
> Jul 04 05:01:26 | *word = E, len = 1
> Jul 04 05:01:26 | *word = 1, len = 1
> Jul 04 05:01:26 | *word = 0, len = 1
> Jul 04 05:01:26 | *word = A, len = 1
> Jul 04 05:01:26 | *word = 0, len = 1
> Jul 04 05:01:26 | alpa-0x000080: rr_weight = 1 (internal default)
> Jul 04 05:01:26 | alpa-0x000080: pgfailback = -2 (config file default)
> Jul 04 05:01:26 | alpa-0x000080: no_path_retry = 3 (config file default)
> Jul 04 05:01:26 | libdevmapper: ioctl/libdm-iface.c(1606): dm message
> alpa-0x000080  NF  queue_if_no_path
>  [16384]
> Jul 04 05:01:26 | pg_timeout = NONE (internal default)
> Jul 04 05:01:26 | alpa-0x000080: event checker started
> Jul 04 05:01:26 | sdaa path added to devmap alpa-0x000080
> Jul 04 05:01:26 | discard event on /block/sdaa/sdaa1
> Segmentation fault
> 
Oh, that one.

Fixed by the attached path.

Christophe, care to apply?

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: Markus Rex, HRB 16746 (AG Nürnberg)

[-- Attachment #2: multipath-tools-uev-discard-segfault --]
[-- Type: text/plain, Size: 761 bytes --]

commit 223885b66311bbb47e439557eced066b3324ad8e
Author: Hannes Reinecke <hare@suse.de>
Date:   Tue Mar 10 08:57:18 2009 +0100

    Stack overflow in uev_trigger()
    
    uev_discard uses sscanf to write a 10 byte string into an array,
    but I forgot to take the trailing NULL byte into account.
    
    References: bnc#476540
    
    Signed-off-by: Hannes Reinecke <hare@suse.de>

diff --git a/multipathd/main.c b/multipathd/main.c
index 7c995b6..1d8ff5a 100644
--- a/multipathd/main.c
+++ b/multipathd/main.c
@@ -652,7 +652,7 @@ uxsock_trigger (char * str, char ** reply, int * len, void * trigger_data)
 static int
 uev_discard(char * devpath)
 {
-	char a[10], b[10], *c;
+	char a[11], b[11], *c;
 
 	/*
 	 * keep only block devices, discard partitions

[-- Attachment #3: Type: text/plain, Size: 0 bytes --]