Skip to contentSkip to conversation

Current git cannot find root device on x86-64 (HP nx6325), NFS broken

2 regressions Rafael J. Wysocki 3 participants 4 messages Apr 6, 2009 – Apr 7, 2009 NFS client

lore
Share & tools
lore
#0 lore raw

Hi,

Current mainline (commit 22ae77bc7ac115b9d518d5cbc13d39317079b2b0) can't find the root device (it's handled by statically compiled-in sata_sil) and generally causes the "invalid module format" message to appear for all modules loaded from the initrd on my HP nx6325 (the distro is openSUSE 11.1).

I don't know which commit might have made this happen, but the breakage was introduced during the last three days (commit 601cc11d054ae4b5e9b5babec3d8e4667a2cb9b5 is certainly good). I'll bisect tomorrow if there are no ideas.

Also, mounting directories over NFS doesn't work (I think it oopses, but didn't have the time to look at it more closely).

Thanks,
Rafael

#1 lore raw
Quoted 2 lines · Rafael J. Wysocki

Also, mounting directories over NFS doesn't work (I think it oopses, but didn't have the time to look at it more closely).

There is a known double free issue that affects the 'proto=' mount option (we call kfree(string) twice). Please could you check that the following patch fixes it for you?

Cheers
Trond

--------------------------------------------------------------------

Quoted 1 line

From a3e6cc62dce11c75249f01423eb9b41531759d36 Mon Sep 17 00:00:00 2001

From: Trond Myklebust <[email protected]>
Date: Mon, 6 Apr 2009 16:41:35 -0700
Subject: [PATCH] NFS: Fix a double free in nfs_parse_mount_options()

Due to an apparent typo, commit a67d18f89f5782806135aad4ee012ff78d45aae7 (NFS: load the rpc/rdma transport module automatically) lead to the 'proto=' mount option doing a double free, while Opt_mountproto leaks a string.

---
 fs/nfs/super.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
fs/nfs/super.c +1 -1
index 82eaadb..6717200 100644
@@ -1228,7 +1228,6 @@ static int nfs_parse_mount_options(char *raw,
 				goto out_nomem;
 			token = match_token(string,
 					    nfs_xprt_protocol_tokens, args);
-			kfree(string);
 
 			switch (token) {
 			case Opt_xprt_udp:
@@ -1258,6 +1257,7 @@ static int nfs_parse_mount_options(char *raw,
 				goto out_nomem;
 			token = match_token(string,
 					    nfs_xprt_protocol_tokens, args);
+			kfree(string);
 
 			switch (token) {
 			case Opt_xprt_udp:
Signature · 11 lines
1.6.0.4



-- 
Trond Myklebust
Linux NFS client maintainer

NetApp
[email protected]
www.netapp.com
#2 lore raw
Quoted 7 lines · Trond Myklebust
Quoted 2 lines · Rafael J. Wysocki

Also, mounting directories over NFS doesn't work (I think it oopses, but didn't have the time to look at it more closely).

There is a known double free issue that affects the 'proto=' mount option (we call kfree(string) twice). Please could you check that the following patch fixes it for you?

The problem is fixed in the current -git, so I guess the patch is in there. :-)

Thanks,
Rafael

#3 lore raw
In reply to Rafael J. Wysocki
Quoted 11 lines · Rafael J. Wysocki

Hi,

Current mainline (commit 22ae77bc7ac115b9d518d5cbc13d39317079b2b0) can't find the root device (it's handled by statically compiled-in sata_sil) and generally causes the "invalid module format" message to appear for all modules loaded from the initrd on my HP nx6325 (the distro is openSUSE 11.1).

I don't know which commit might have made this happen, but the breakage was introduced during the last three days (commit 601cc11d054ae4b5e9b5babec3d8e4667a2cb9b5 is certainly good). I'll bisect tomorrow if there are no ideas.

I bisected "invalid module format" woes to 9cb610d.

-Mike