Re: linux-next: Tree for Jan 7 (microblaze qemu problems)

[Guenter Roeck]

On Wed, Jan 07, 2015 at 01:45:30PM -0800, Guenter Roeck wrote:
> On Wed, Jan 07, 2015 at 03:16:18PM +1100, Stephen Rothwell wrote:
> > Hi all,
> > 
> > Changes since 20150106:
> > 
> > *crickets*
> > 
> > Non-merge commits (relative to Linus' tree): 1350
> >  1543 files changed, 41856 insertions(+), 24250 deletions(-)
> > 
> 
> With this kernel I get the following WARNINGs (lots of htem) when running
> microblaze images (both big and little endian) with qemu.
> 
> WARNING: CPU: 0 PID: 1 at arch/microblaze/kernel/cpu/cache.c:432
> __invalidate_dcache_range_msr_irq_wt+0xcc/0xd8()
> Modules linked in:
> CPU: 0 PID: 1 Comm: swapper Not tainted 3.19.0-rc3-next-20150107 #1
> ...
> (sample) Call Trace:
> [<c0004080>] microblaze_unwind+0x5c/0x78
> [<c0003d64>] show_stack+0xe0/0x164
> [<c02e2224>] dump_stack+0x20/0x38
> [<c000adbc>] warn_slowpath_common+0x90/0xec
> [<c000aea0>] warn_slowpath_null+0xc/0x24
> [<c000493c>] __invalidate_dcache_range_msr_irq_wt+0xc0/0xd8
> [<c00c59c4>] simple_write_begin+0xf8/0x170
> [<c005d18c>] generic_perform_write+0xa8/0x2ac
> [<c005f1f8>] __generic_file_write_iter+0x1a8/0x478
> [<c005f504>] generic_file_write_iter+0x3c/0x128
> [<c009b0c4>] new_sync_write+0x84/0xcc
> [<c009ba88>] vfs_write+0xe8/0x224
> [<c009c314>] SyS_write+0x4c/0xf4
> [<c03d5200>] xwrite+0x48/0xb8
> [<c03d52a0>] do_copy+0x30/0x1ac
> [<c03d47b4>] write_buffer+0x34/0x60
> [<c03d4810>] flush_buffer+0x30/0xd0
> [<c03e4c70>] gunzip+0x2c0/0x3ac
> [<c03d4c20>] unpack_to_rootfs+0x1b8/0x368
> [<c03d56e8>] populate_rootfs+0x28/0x2d8
> [<c000044c>] do_one_initcall+0xa0/0x24c
> [<c03d2d9c>] kernel_init_freeable+0x150/0x238
> [<c02dedf0>] kernel_init+0x8/0x14c
> [<c0005d78>] ret_from_kernel_thread+0x8/0x20
> 
> Detailed logs are available at http://server.roeck-us.net:8010/builders.
> 
> bisect results:
> 
> # bad: [7e3619a6de57f0257a2f6480182d2287ee05e314] Add linux-next specific files for 20150107
> # good: [b1940cd21c0f4abdce101253e860feff547291b0] Linux 3.19-rc3
> git bisect start 'HEAD' 'v3.19-rc3'
> # bad: [c48667c5659248ff2b2fbff5cd23c43b5768e81b] Merge remote-tracking branch 'net-next/master'
> git bisect bad c48667c5659248ff2b2fbff5cd23c43b5768e81b
> # bad: [5aa2843082a0ce9b710549bdc7ae8d54140edc70] Merge remote-tracking branch 'pci/next'
> git bisect bad 5aa2843082a0ce9b710549bdc7ae8d54140edc70
> # good: [11339e9e3d575c1486ff57dc49c83b6154918305] Merge remote-tracking branch 'samsung/for-next'
> git bisect good 11339e9e3d575c1486ff57dc49c83b6154918305
> # bad: [50d16e144b253542d89830c0ace8573a67ce4b16] Merge remote-tracking branch 'powerpc-mpe/next'
> git bisect bad 50d16e144b253542d89830c0ace8573a67ce4b16
> # good: [3fda5c0dda2eafe244898809a74831209cdb69b7] Merge remote-tracking branch 'cris/for-next'
> git bisect good 3fda5c0dda2eafe244898809a74831209cdb69b7
> # bad: [8b772494fda25dd998ade5621cc0684a5c897179] Merge remote-tracking branch 'microblaze/next'
> git bisect bad 8b772494fda25dd998ade5621cc0684a5c897179
> # bad: [cd2b00db005541d15ea9fbb474b5e5e98b993d8e] microblaze: Fix variable types to remove W=1 warning
> git bisect bad cd2b00db005541d15ea9fbb474b5e5e98b993d8e
> # good: [8543e6c96762fcc930af5725088c2b9e4865c3aa] microblaze: Use unsigned return type in do_syscall_trace_enter
> git bisect good 8543e6c96762fcc930af5725088c2b9e4865c3aa
> # good: [e76fdb324844cc0367f8c6ea5d13278a81568ccb] microblaze: Use unsigned type for "for" loop because of comparison-kgdb.c
> git bisect good e76fdb324844cc0367f8c6ea5d13278a81568ccb
> # bad: [b343c7d621bf801b502cba812c3dcb4320942b93] microblaze: Use unsigned type for limit comparison in cache.c
> git bisect bad b343c7d621bf801b502cba812c3dcb4320942b93
> # good: [bdb96e3cad21f5973c95f0e6687db4a57eff7c53] microblaze: Use unsigned type for proper comparison in cpuinfo*.c
> git bisect good bdb96e3cad21f5973c95f0e6687db4a57eff7c53
> # first bad commit: [b343c7d621bf801b502cba812c3dcb4320942b93] microblaze: Use unsigned type for limit comparison in cache.c
> 
> Reverting the offending commit fixes the problem.
> 
Followup: I think the following should fix the problem.
My qemu tests pass with this change.

Original patch:

-	WARN_ON(end - start < 0);                                       \
+	WARN_ON(end > start);                                   \

This looks wrong, since 'end - start < 0' is equivalent to 'end < start',
not 'end > start'.

Guenter

---
diff --git a/arch/microblaze/kernel/cpu/cache.c b/arch/microblaze/kernel/cpu/cache.c
index 84c9558..0bde47e 100644
--- a/arch/microblaze/kernel/cpu/cache.c
+++ b/arch/microblaze/kernel/cpu/cache.c
@@ -143,7 +143,7 @@ do {									\
 	unsigned int volatile temp = 0;						\
 	unsigned int align = ~(line_length - 1);					\
 	end = ((end & align) == end) ? end - line_length : end & align;	\
-	WARN_ON(end > start);					\
+	WARN_ON(end < start);					\
 									\
 	__asm__ __volatile__ (" 1:	" #op "	%1, r0;"		\
 					"cmpu	%0, %1, %2;"		\
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/