Re: making reiser4/AMD64 hardlock

[Vladimir Saveliev <vs@namesys.com>]

[-- Attachment #1: Type: text/plain, Size: 2132 bytes --]

Hello

On Wed, 2004-11-10 at 11:01, Jake Maciejewski wrote:
> I documented a few more AMD64 errors/panics. The system hasn't been
> freezing since I enabled debugging, but make, dd, or whatever else hits
> the bug(s) still freeze.
> 
> http://people.msoe.edu/~maciejej/patches/AMD64_reiser4_debug/11-08-04/
> with reiser4progs 1.0.2 and my custom patched 2.6.9 kernel
> 
> http://people.msoe.edu/~maciejej/patches/AMD64_reiser4_debug/11-09-04/
> with reiser4progs 1.0.2 and 2.6.10-rc1 with the ftp.namesys.com patch
> 

Would you please try to reproduce this problem (reiser4[dd(7926)]:
check_blocks_bitmap (fs/reiser4/plugin/space/bitmap.c:1174)[zam-623])
with the attached patch? 

> Every test has been on a freshly made filesystem.
> 
> You might be interested to know that I think fsck failed to fix
> corruption after my 11-08-04/test1. When I tried to build the kernel
> after a --build-fs, make failed. I didn't have checksums to verify the
> tree, so something else could have been wrong. It froze when I tried to
> dump metadata.
> 
> On Thu, 2004-11-04 at 12:52 +0300, Vladimir Saveliev wrote:
> > Hello
> > 
> > On Wed, 2004-11-03 at 23:18, Hendrik Visage wrote:
> > > On Wed, Nov 03, 2004 at 02:59:19AM -0600, Jake Maciejewski wrote:
> > > > I've been testing reiser4 on 2.6.9 (patches from 2.6.9-mm1 and fixes
> > > > from reiser4-for-2.6.9.diff). I have reiser4progs and libaal 1.0.1.
> > > > Syslog doesn't catch any errors when I get hardlocks (haven't tried
> > > > SysRq). I figured I could at least give you guys a hint about what kind
> > > > of usage pattern kills reiser4.
> > > 
> > 
> > Please try to get as much debugging information as you can.
> > sysrq+t's output may help to understand the problem. Do you have "File
> > systems" -> "Reiser4" -> "Enable reiser4 debug options" -> "Assertions"
> > turned on? If no, please turn it, it may also help. Try to catch its
> > output, via serial console if it will not be stored in logs.
> > 
> > I will try your test in x86.
> > 
> > > I recall the last response about this issue:
> > > 
> > >  We need an AMD64 cpu...
> > > 
> > well, yes.
> > 
> > > 
> > 

[-- Attachment #2: bitmap.c.diff --]
[-- Type: text/plain, Size: 1184 bytes --]

--- reiser4.orig/plugin/space/bitmap.c	2004-11-09 16:30:37.991446947 +0300
+++ reiser4/plugin/space/bitmap.c	2004-11-10 19:01:18.450540361 +0300
@@ -337,6 +337,7 @@ reiser4_find_last_zero_bit (bmap_off_t *
 static void
 reiser4_clear_bits(char *addr, bmap_off_t start, bmap_off_t end)
 {
+/*
 	int first_byte;
 	int last_byte;
 
@@ -360,6 +361,14 @@ reiser4_clear_bits(char *addr, bmap_off_
 		addr[first_byte] &= first_byte_mask;
 		addr[last_byte] &= last_byte_mask;
 	}
+*/
+	int i;
+	bmap_off_t count;
+
+	BUG_ON(end <= start);
+	count = end - start;
+	for (i = 0; i < count; i ++)
+		reiser4_clear_bit(start + i, addr);
 }
 
 /* Audited by: green(2002.06.12) */
@@ -367,6 +376,7 @@ reiser4_clear_bits(char *addr, bmap_off_
 static void
 reiser4_set_bits(char *addr, bmap_off_t start, bmap_off_t end)
 {
+#if 0
 	int first_byte;
 	int last_byte;
 
@@ -390,6 +400,14 @@ reiser4_set_bits(char *addr, bmap_off_t 
 		addr[first_byte] |= first_byte_mask;
 		addr[last_byte] |= last_byte_mask;
 	}
+#endif
+	int i;
+	bmap_off_t count;
+
+	BUG_ON(end <= start);
+	count = end - start;
+	for (i = 0; i < count; i ++)
+		reiser4_set_bit(start + i, addr);
 }
 
 #define ADLER_BASE    65521