Re: 2.6.19-rc6-mm1

[Andy Whitcroft <apw@shadowen.org>]

Andrew Morton wrote:
> On Thu, 23 Nov 2006 12:23:48 +0100
> Mariusz Kozlowski <m.kozlowski@tuxland.pl> wrote:
> 
>> Hello,
>>
>> 	Hmmm ... didn't apply cleanly.
>>
>> patching file kernel/tsacct.c
>> Hunk #1 FAILED at 97.
>> 1 out of 1 hunk FAILED -- saving rejects to file kernel/tsacct.c.rej
> 
> I think your local tree is not clean.

I get this accross the board on my test system too.  All clean downloads.

A quick look at the combo-patch and the broken-out patch seems to
indicate they are not in sync with each other.  In the combo-patch we
have this hunk (which is the one which fails):

--- linux-2.6.19-rc6/kernel/tsacct.c    2006-11-16 23:19:32.000000000 -0800
+++ devel/kernel/tsacct.c       2006-11-23 01:12:17.000000000 -0800
@@ -97,7 +97,14 @@ void xacct_add_tsk(struct taskstats *sta
        stats->read_syscalls    = p->syscr;
        stats->write_syscalls   = p->syscw;
 #ifdef CONFIG_TASK_IO_ACCOUNTING
-       stats->read_bytes       = p->ioac->read_bytes
+       stats->read_bytes       = p->ioac.read_bytes;
+       stats->write_bytes      = p->ioac.write_bytes;
+       stats->cancelled_write_bytes = p->ioac.cancelled_write_bytes;
+#else
+       stats->read_bytes       = 0;
+       stats->write_bytes      = 0;
+       stats->cancelled_write_bytes = 0;
+#endif

In the broken-out directory the only patch which references this file
has the following different hunk:

--- a/kernel/tsacct.c~io-accounting-via-taskstats
+++ a/kernel/tsacct.c
@@ -96,6 +96,15 @@ void xacct_add_tsk(struct taskstats *sta
        stats->write_char       = 0;
        stats->read_syscalls    = p->syscr;
        stats->write_syscalls   = p->syscw;
+#ifdef CONFIG_TASK_IO_ACCOUNTING
+       stats->read_bytes       = p->ioac.read_bytes;
+       stats->write_bytes      = p->ioac.write_bytes;
+       stats->cancelled_write_bytes = p->ioac.cancelled_write_bytes;
+#else
+       stats->read_bytes       = 0;
+       stats->write_bytes      = 0;
+       stats->cancelled_write_bytes = 0;
+#endif
 }
 #undef KB
 #undef MB

Looking at 2.6.19-rc6 this second version seems completely reasonable.
The former does not.

-apw