Skip to contentSkip to conversation

cifs hang patch idea - reduce sendtimeo on socket

DISCUSSION Dave Kleikamp 3 participants 3 messages Feb 22, 2006 – Feb 26, 2006 2 messages not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#2 lore raw
Quoted 3 lines · Adrian Bunk

I'm a bit lost now, but I hope this information helps you in finding what is going wrong?

Steve and I think we have figured this out. In some cases, CIFSSMBRead was returning a recently freed buffer.

CIFS: CIFSSMBRead was returning an invalid pointer in buf

Thanks to Adrian Bunk for debugging the problem

Also added a fix for 64K pages we found in loosely-related testing

fs/cifs/cifssmb.c +4 -3
index 38ab9f6..9d7bbd2 100644
@@ -1076,13 +1076,14 @@ CIFSSMBRead(const int xid, struct cifsTc
 			cifs_small_buf_release(iov[0].iov_base);
 		else if(resp_buf_type == CIFS_LARGE_BUFFER)
 			cifs_buf_release(iov[0].iov_base);
-	} else /* return buffer to caller to free */ /* BB FIXME how do we tell caller if it is not a large buffer */ {
-		*buf = iov[0].iov_base;
+	} else if(resp_buf_type != CIFS_NO_BUFFER) {
+		/* return buffer to caller to free */ 
+		*buf = iov[0].iov_base;		
 		if(resp_buf_type == CIFS_SMALL_BUFFER)
 			*pbuf_type = CIFS_SMALL_BUFFER;
 		else if(resp_buf_type == CIFS_LARGE_BUFFER)
 			*pbuf_type = CIFS_LARGE_BUFFER;
-	}
+	} /* else no valid buffer on return - leave as null */
 
 	/* Note: On -EAGAIN error only caller can retry on handle based calls
 		since file handle passed in no longer valid */
fs/cifs/connect.c +4 -4
index 0e1560a..16535b5 100644
@@ -1795,10 +1795,10 @@ cifs_mount(struct super_block *sb, struc
 			   conjunction with 52K kvec constraint on arch with 4K
 			   page size  */
 
-		if(cifs_sb->rsize < PAGE_CACHE_SIZE) {
-			cifs_sb->rsize = PAGE_CACHE_SIZE; 
-			/* Windows ME does this */
-			cFYI(1,("Attempt to set readsize for mount to less than one page (4096)"));
+		if(cifs_sb->rsize < 2048) {
+			cifs_sb->rsize = 2048; 
+			/* Windows ME may prefer this */
+			cFYI(1,("readsize set to minimum 2048"));
 		}
 		cifs_sb->mnt_uid = volume_info.linux_uid;
 		cifs_sb->mnt_gid = volume_info.linux_gid;
Signature · 2 lines
David Kleikamp
IBM Linux Technology Center
#3 lore raw
Quoted 8 lines · Dave Kleikamp
Quoted 3 lines · Adrian Bunk

I'm a bit lost now, but I hope this information helps you in finding what is going wrong?

Steve and I think we have figured this out. In some cases, CIFSSMBRead was returning a recently freed buffer.
...

Thanks a lot!

In my testing, this patch applied against 2.6.14-rc4 fixed all problems I observed.

Quoted 1 line

David Kleikamp

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#4 lore raw
Quoted 11 lines · Adrian Bunk
Quoted 5 lines

Steve and I think we have figured this out. In some cases, CIFSSMBRead was returning a recently freed buffer.
...

Thanks a lot!

In my testing, this patch applied against 2.6.14-rc4 fixed all problems I observed.

Adrian,
Thanks for the info. Our testing looked good too.

In the development tree for cifs we are working on reducing buffer usage (a good first step is already checked in) and enabling additional important security features (NTLMv2 and Kerberos for newer more secure clients and also an option to configure and mount for allowing lanman and plaintext sessionsetup to allow mounts to older legacy servers).