Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Please excuse my previous email with no details.... here is a patch that
addresses the first issue I am seeing on the ARM platform.
This addresses the issue of the 'Unknown tag' error. As you suspected
this is an alignment issue, if the data (in my case /dev/loop0 + NULL
byte) has an odd size, the previous code failed to store the tag, for
the subsequent parameter, due to the data pointer not falling on a word
boundary. This fixes it for ARM, and it continues to work on i386,
however I am not sure about other platforms, perhaps this would be an
issue on big endian platforms, since the add_tag function takes the tag
and size as an int, rather than a short.
Hopefully this will be the only issue I have!
Regards
Nick
-- drbdsetup.c.orig 2009-06-11 10:51:08.000000000 +0100
+++ drbdsetup.c 2009-06-11 10:53:43.000000000 +0100
@@ -549,8 +549,8 @@
fprintf(stderr, "Tag list size exceeded!\n");
exit(20);
}
- *tl->tag_list_cpos++ = tag;
- *tl->tag_list_cpos++ = data_len;
+ memcpy(tl->tag_list_cpos++, (short*)&tag, sizeof(tl->tag_list_cpos));
+ memcpy(tl->tag_list_cpos++, (short*)&data_len,
sizeof(tl->tag_list_cpos));
memcpy(tl->tag_list_cpos,data,data_len);
tl->tag_list_cpos = (unsigned short*)((char*)tl->tag_list_cpos +
data_len);
}
Lars Ellenberg wrote:
> On Wed, Jun 10, 2009 at 02:04:06PM +0100, Nick Liebmann wrote:
>
>> Thanks for the speedy response Lars, I have patched my kernel module
>> (its 8.3.1, so I just changed the kmalloc to a kzalloc...I hope that is
>> OK), and now no longer suffer from the kernel Oops.. great, but
>> unfortunately I am still getting the REAL problem
>>
>> kernel: drbd0: Unknown tag: 1071
>>
>> nasty:~# drbdadm attach r0
>> 0: Failure: (126) UnknownMandatoryTag
>> Command 'drbdsetup 0 disk /dev/loop0 /dev/loop0 internal --set-defaults --create-device' terminated with exit code 10
>>
>>
>> I have found no reference to this on the mailing lists, any ideas what is going on here?
>>
>
> nope.
> maybe alignment issues.
> maybe a build problem.
> you need to debug that.
>
>