connectathon special test failure in recent kernels

Chuck Lever

author

2006-08-08 19:20 UTC #0 lore raw

Is the connectathon holey test failure going to be fixed before 2.6.18 is finalized?

This is from a 2.6.18-rc3 client:

test exclusive create.

test negative seek, you should get: read: Invalid argument or lseek: Invalid argument
lseek: Invalid argument

test rename

test truncate
truncate succeeded

test holey file support
read (hole) offset 8192, sz = 56667, bytes = 5141 (ret -1), holesz = 9012
read: Input/output error
special tests failed
Tests failed, leaving /mnt/brahms mounted
[cel@picasso cthon04]$

Signature · 11 lines

"We who cut mere stones must always be envisioning cathedrals"
   -- Quarry worker's creed

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist  -  [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Trond Myklebust

2006-08-08 23:33 UTC #1 lore raw

Quoted 2 lines · Chuck Lever

Is the connectathon holey test failure going to be fixed before 2.6.18 is finalized?

Depends. Have you traced it in order to figure out what is wrong?

Cheers,
Trond

Quoted 13 lines

This is from a 2.6.18-rc3 client:
test exclusive create.
test negative seek, you should get: read: Invalid argument or lseek: Invalid argument
lseek: Invalid argument
test rename
test truncate
truncate succeeded

Quoted 7 lines

test holey file support
read (hole) offset 8192, sz = 56667, bytes = 5141 (ret -1), holesz = 9012
read: Input/output error
special tests failed
Tests failed, leaving /mnt/brahms mounted
[cel@picasso cthon04]$

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Chuck Lever

author

2006-08-09 02:59 UTC #2 lore raw

Quoted 5 lines · Trond Myklebust

Quoted 2 lines · Chuck Lever
Is the connectathon holey test failure going to be fixed before 2.6.18 is finalized?
Depends. Have you traced it in order to figure out what is wrong?

Ah. Bruce reported this problem a while back, and said that it goes away if one of your recent patches to fs/nfs/read.c is reverted. Bruce?

Signature · 11 lines

"We who cut mere stones must always be envisioning cathedrals"
   -- Quarry worker's creed

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist  -  [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Steve Dickson

2006-08-17 15:10 UTC #3 lore raw

Quoted 7 lines · Chuck Lever

Quoted 5 lines · Trond Myklebust
Quoted 2 lines · Chuck Lever
Is the connectathon holey test failure going to be fixed before 2.6.18 is finalized?

Quoted 6 lines

Quoted 1 line
Depends. Have you traced it in order to figure out what is wrong?
Ah. Bruce reported this problem a while back, and said that it goes away if one of your recent patches to fs/nfs/read.c is reverted. Bruce?

Well the patch in question is:

commit 1de3fc12ea085690547a54b6efa01c7348f1cebd
tree ea865786120cfcefac563c54693fef8d3d718f10
parent 128e6ced247cda88f96fa9f2e4ba8b2c4a681560
author Trond Myklebust <[email protected]> Thu, 25 May 2006 09:40:44 -0400
committer Trond Myklebust <[email protected]> Fri, 09 Jun 2006 17:34:03 -0400

NFS: Clean up and fix page zeroing when we have short reads

The code that is supposed to zero the uninitialised partial pages when the server returns a short read is currently broken: it looks at the nfs_page wb_pgbase and wb_bytes fields instead of the equivalent nfs_read_data values when deciding where to start truncating the page.

Also ensure that we are more careful about setting PG_uptodate before retrying a short read: the retry will change the nfs_read_data args.pgbase and args.count.

I'm looking into now... but this patch definitely breaks the holy test...

steved.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Chuck Lever

author

2006-08-17 16:40 UTC #4 lore raw

Steve-

For the record, you did mention to me at one point that you saw this against Linux *and* Solaris NFS servers, right?

Quoted 52 lines · Steve Dickson

Quoted 7 lines · Chuck Lever
Quoted 5 lines · Trond Myklebust
Quoted 2 lines · Chuck Lever
Is the connectathon holey test failure going to be fixed before 2.6.18 is finalized?
Quoted 6 lines
Quoted 1 line
Depends. Have you traced it in order to figure out what is wrong?
Ah. Bruce reported this problem a while back, and said that it goes away if one of your recent patches to fs/nfs/read.c is reverted. Bruce?
Well the patch in question is:
commit 1de3fc12ea085690547a54b6efa01c7348f1cebd
tree ea865786120cfcefac563c54693fef8d3d718f10
parent 128e6ced247cda88f96fa9f2e4ba8b2c4a681560
author Trond Myklebust <[email protected]> Thu, 25 May 2006 09:40:44 -0400
committer Trond Myklebust <[email protected]> Fri, 09 Jun 2006 17:34:03 -0400
NFS: Clean up and fix page zeroing when we have short reads
The code that is supposed to zero the uninitialised partial pages when the server returns a short read is currently broken: it looks at the nfs_page wb_pgbase and wb_bytes fields instead of the equivalent nfs_read_data values when deciding where to start truncating the page.
Also ensure that we are more careful about setting PG_uptodate before retrying a short read: the retry will change the nfs_read_data args.pgbase and args.count.
I'm looking into now... but this patch definitely breaks the holy test...
steved.
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Signature · 11 lines

"We who cut mere stones must always be envisioning cathedrals"
   -- Quarry worker's creed

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist  -  [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Steve Dickson

2006-08-17 17:07 UTC #5 lore raw

Quoted 4 lines · Chuck Lever

Steve-
For the record, you did mention to me at one point that you saw this against Linux *and* Solaris NFS servers, right?

This is true.... This failure happens against both flavors...

steved.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Trond Myklebust

2006-08-17 19:54 UTC #6 lore raw

Quoted 7 lines · Steve Dickson

Quoted 4 lines · Chuck Lever
Steve-
For the record, you did mention to me at one point that you saw this against Linux *and* Solaris NFS servers, right?
This is true.... This failure happens against both flavors...

Could you send us an example of 'fsx' output that demonstrates the problem?

Cheers,
Trond

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Steve Dickson

2006-08-17 20:54 UTC #7 lore raw

Quoted 3 lines · Trond Myklebust

Could you send us an example of 'fsx' output that demonstrates the problem?

hmmm... fsx? What does the fsx have to do with the special tests of the connectathon failing... I guess I'm missing something... anyways... I'll try to run some fsx test runs asaic...

steved.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Trond Myklebust

2006-08-17 22:37 UTC #8 lore raw

Quoted 7 lines · Steve Dickson

Quoted 3 lines · Trond Myklebust
Could you send us an example of 'fsx' output that demonstrates the problem?
hmmm... fsx? What does the fsx have to do with the special tests of the connectathon failing... I guess I'm missing something... anyways... I'll try to run some fsx test runs asaic...

fsx creates holes in files left, right and center, and then tests for correctness. If there is a problem with the read patch, and hole creation, then fsx should find it and provide you with a nice dump of what went wrong, where, and what sequence of events lead up to the problem.

Cheers,
Trond

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

J. Bruce Fields

2006-08-18 02:50 UTC #9 lore raw

Quoted 9 lines · Trond Myklebust

Quoted 7 lines · Steve Dickson
Quoted 3 lines · Trond Myklebust
Could you send us an example of 'fsx' output that demonstrates the problem?
hmmm... fsx? What does the fsx have to do with the special tests of the connectathon failing... I guess I'm missing something... anyways... I'll try to run some fsx test runs asaic...

Quoted 5 lines

fsx creates holes in files left, right and center, and then tests for correctness. If there is a problem with the read patch, and hole creation, then fsx should find it and provide you with a nice dump of what went wrong, where, and what sequence of events lead up to the problem.

The holey test run on its own also gives a pretty short simple test case.--b.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Trond Myklebust

2006-08-18 02:55 UTC #10 lore raw

Quoted 2 lines · J. Bruce Fields

The holey test run on its own also gives a pretty short simple test case.--b.

It would if it actually reproduced the problem on my setup. As it doesn't, fsx is the second best option since it produces a full dump of the offending file section.

Trond

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

Steve Dickson

2006-08-21 02:05 UTC #11 lore raw

PATCH 1/1 connectathon special test failure in recent kernels

In reply to Steve Dickson

Quoted 11 lines · Steve Dickson

Well the patch in question is:
commit 1de3fc12ea085690547a54b6efa01c7348f1cebd
tree ea865786120cfcefac563c54693fef8d3d718f10
parent 128e6ced247cda88f96fa9f2e4ba8b2c4a681560
author Trond Myklebust <[email protected]> Thu, 25 May 2006 09:40:44 -0400
committer Trond Myklebust <[email protected]> Fri, 09 Jun 2006 17:34:03 -0400
NFS: Clean up and fix page zeroing when we have short reads

Attached is that patch that fixes the regression caused by the above patch... I've tested with both the connectathon test suite and the fsx test suite.

The patch is based on the 2.6.18-rc4 kernel tree.

steved.

The patch fixes a regression that causes the special tests of the Connectation tests suite to fail. When a hole in a file is created (like the holey test does), the page will be allocated but there will be nothing read into it. The means the number of byte read from the server will be zero (i.e. data->res.count). In the case, the page still has PageUptodate-ed which is what this patch does.

Signed-off-by Steve Dickson <[email protected]>

fs/nfs/read.c renamed from fs/nfs/read.c.orig +4 -3

@@ -476,13 +476,14 @@
 	unsigned int base = data->args.pgbase;
 	struct page **pages;
 
-	if (unlikely(count == 0))
-		return;
 	pages = &data->args.pages[base >> PAGE_CACHE_SHIFT];
 	base &= ~PAGE_CACHE_MASK;
 	count += base;
-	for (;count >= PAGE_CACHE_SIZE; count -= PAGE_CACHE_SIZE, pages++)
+	if (count == 0) 
 		SetPageUptodate(*pages);
+	else 
+		for (;count >= PAGE_CACHE_SIZE; count -= PAGE_CACHE_SIZE, pages++)
+			SetPageUptodate(*pages);
 	/*
 	 * Was this an eof or a short read? If the latter, don't mark the page
 	 * as uptodate yet.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs