Re: [patch/rfc] KFAIL EBUSY kernel bug

Re: [patch/rfc] KFAIL EBUSY kernel bug

[Michael Elizabeth Chastain]

> diff -u -r1.5 fileio.exp
> --- gdb.base/fileio.exp	10 Nov 2003 09:12:20 -0000	1.5
> +++ gdb.base/fileio.exp	8 Jan 2004 20:09:20 -0000

Yo!  Copyright police!

I proof-read the rest of the patch and it looks stylin' to me.
gdb_test_multiple rocks.  The comment on the kfail is cool.

Michael C

[patch/rfc] KFAIL EBUSY kernel bug

[Andrew Cagney]

[-- Attachment #1: Type: text/plain, Size: 211 bytes --]

Hello,

Ref: http://sources.redhat.com/gdb/bugs/1502
Turns out that the EBUSY error status value I was seeing is a Linux 
Kernel bug.  This KFAIL's that case.

baring problems, I'll commit in a few days

Andrew

[-- Attachment #2: diffs --]
[-- Type: text/plain, Size: 1327 bytes --]

2004-01-08  Andrew Cagney  <cagney@redhat.com>

	* gdb.base/fileio.exp: For "Renaming a directory to a non-empty
	directory returns ENOTEMPTY or EEXIST", treat EBUSY as a KFAIL.

Index: gdb.base/fileio.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/fileio.exp,v
retrieving revision 1.5
diff -u -r1.5 fileio.exp
--- gdb.base/fileio.exp	10 Nov 2003 09:12:20 -0000	1.5
+++ gdb.base/fileio.exp	8 Jan 2004 20:09:20 -0000
@@ -216,9 +216,16 @@
 "Renaming a file to existing directory returns EISDIR"
 
 send_gdb "tbreak 388\n" ; gdb_expect -re "$gdb_prompt $"
-gdb_test continue \
-	"Continuing\\..*rename 3:.*(ENOTEMPTY|EEXIST).*test_rename \\(\\) at.*$srcfile:388.*" \
-"Renaming a directory to a non-empty directory returns ENOTEMPTY or EEXIST"
+set test "Renaming a directory to a non-empty directory returns ENOTEMPTY or EEXIST"
+gdb_test_multiple continue "${test}" {
+    -re "Continuing\\..*rename 3:.*(ENOTEMPTY|EEXIST).*test_rename \\(\\) at.*$srcfile:388.*" {
+	pass "${test}"
+    }
+    -re "Continuing\\..*rename 3:.*EBUSY.*test_rename \\(\\) at.*$srcfile:388.*" {
+	# On NFS, GNU/Linux <= 2.6/2004-01-08 at least, gets this wrong
+	kfail gdb/1502 "${test}"
+    }
+}
 
 send_gdb "tbreak 393\n" ; gdb_expect -re "$gdb_prompt $"
 gdb_test continue \

Re: [patch/rfc] KFAIL EBUSY kernel bug

[Daniel Jacobowitz]

On Thu, Jan 08, 2004 at 03:12:23PM -0500, Andrew Cagney wrote:
> Hello,
> 
> Ref: http://sources.redhat.com/gdb/bugs/1502
> Turns out that the EBUSY error status value I was seeing is a Linux 
> Kernel bug.  This KFAIL's that case.
> 
> baring problems, I'll commit in a few days

If it's a Linux kernel bug, that makes it an XFAIL, not a KFAIL.

> 2004-01-08  Andrew Cagney  <cagney@redhat.com>
> 
> 	* gdb.base/fileio.exp: For "Renaming a directory to a non-empty
> 	directory returns ENOTEMPTY or EEXIST", treat EBUSY as a KFAIL.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [patch/rfc] KFAIL EBUSY kernel bug

[Andrew Cagney]

> On Thu, Jan 08, 2004 at 03:12:23PM -0500, Andrew Cagney wrote:
> 
>> Hello,
>> 
>> Ref: http://sources.redhat.com/gdb/bugs/1502
>> Turns out that the EBUSY error status value I was seeing is a Linux 
>> Kernel bug.  This KFAIL's that case.
>> 
>> baring problems, I'll commit in a few days
> 
> 
> If it's a Linux kernel bug, that makes it an XFAIL, not a KFAIL.

Oops, yes (I had my Red Hat kernel, rather than FSF GDB glasses on).

>> 2004-01-08  Andrew Cagney  <cagney@redhat.com>
>> 
>> 	* gdb.base/fileio.exp: For "Renaming a directory to a non-empty
>> 	directory returns ENOTEMPTY or EEXIST", treat EBUSY as a KFAIL.
> 
> 
> -- Daniel Jacobowitz MontaVista Software Debian GNU/Linux Developer 

Re: [patch/rfc] KFAIL EBUSY kernel bug

[Andrew Cagney]

[-- Attachment #1: Type: text/plain, Size: 673 bytes --]

> On Thu, Jan 08, 2004 at 03:12:23PM -0500, Andrew Cagney wrote:
> 
> Hello,
> 
> Ref: http://sources.redhat.com/gdb/bugs/1502
> Turns out that the EBUSY error status value I was seeing is a Linux Kernel bug.  This KFAIL's that case.
> 
> baring problems, I'll commit in a few days
> 
> 
> If it's a Linux kernel bug, that makes it an XFAIL, not a KFAIL.
> 
> Oops, yes (I had my Red Hat kernel, rather than FSF GDB glasses on).
> 
> 2004-01-08  Andrew Cagney  <cagney@redhat.com>
> 
>     * gdb.base/fileio.exp: For "Renaming a directory to a non-empty
>     directory returns ENOTEMPTY or EEXIST", treat EBUSY as a KFAIL.

Attatched is what I've just committed.

Andrew


[-- Attachment #2: diffs --]
[-- Type: text/plain, Size: 1649 bytes --]

2004-01-12  Andrew Cagney  <cagney@redhat.com>

	* gdb.base/fileio.exp: Update copyright year.  On GNU/Linux, for
	"Renaming a directory to a non-empty directory returns ENOTEMPTY
	or EEXIST", treat EBUSY as an XFAIL.

Index: gdb.base/fileio.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/fileio.exp,v
retrieving revision 1.5
diff -u -r1.5 fileio.exp
--- gdb.base/fileio.exp	10 Nov 2003 09:12:20 -0000	1.5
+++ gdb.base/fileio.exp	12 Jan 2004 14:43:49 -0000
@@ -1,4 +1,4 @@
-#   Copyright 2002, 2003
+#   Copyright 2002, 2003, 2004
 #   Free Software Foundation, Inc.
 
 # This program is free software; you can redistribute it and/or modify
@@ -216,9 +216,19 @@
 "Renaming a file to existing directory returns EISDIR"
 
 send_gdb "tbreak 388\n" ; gdb_expect -re "$gdb_prompt $"
-gdb_test continue \
-	"Continuing\\..*rename 3:.*(ENOTEMPTY|EEXIST).*test_rename \\(\\) at.*$srcfile:388.*" \
-"Renaming a directory to a non-empty directory returns ENOTEMPTY or EEXIST"
+set test "Renaming a directory to a non-empty directory returns ENOTEMPTY or EEXIST"
+gdb_test_multiple continue "${test}" {
+    -re "Continuing\\..*rename 3:.*(ENOTEMPTY|EEXIST).*test_rename \\(\\) at.*$srcfile:388.*" {
+	pass "${test}"
+    }
+    -re "Continuing\\..*rename 3:.*EBUSY.*test_rename \\(\\) at.*$srcfile:388.*" {
+	# At least version <= 2.6/2004-01-08 of the Linux Kernel gets
+	# this wrong (reporting EBUSY) when the file system is NFS
+	# mounted.
+	setup_xfail *-*-*linux* gdb/1502
+	fail "${test}"
+    }
+}
 
 send_gdb "tbreak 393\n" ; gdb_expect -re "$gdb_prompt $"
 gdb_test continue \