Re: kmemleak for MIPS

[Catalin Marinas]

On Mon, 2011-03-28 at 22:15 +0100, Maxin John wrote:
> > Just add "depends on MIPS" and give it a try.
> As per your suggestion, I have tried it in my qemu environment (MIPS malta).
> 
> With a minor modification in arch/mips/kernel/vmlinux.lds.S (added the
> symbol  _sdata ), I was able to add kmemleak support for MIPS.
> 
> Output in MIPS (Malta):

You may want to disable the kmemleak testing to reduce the amount of
leaks reported.

> debian-mips:~# uname -a
> Linux debian-mips 2.6.38-08826-g1788c20-dirty #4 SMP Mon Mar 28
> 23:22:04 EEST 2011 mips GNU/Linux
> debian-mips:~# mount -t debugfs nodev /sys/kernel/debug/
> debian-mips:~# cat /sys/kernel/debug/kmemleak
> unreferenced object 0x8f95d000 (size 4096):
>   comm "swapper", pid 1, jiffies 4294937330 (age 467.240s)
>   hex dump (first 32 bytes):
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>   backtrace:
>     [<80529644>] alloc_large_system_hash+0x2f8/0x410
>     [<8053864c>] udp_table_init+0x4c/0x158
>     [<80538774>] udp_init+0x1c/0x94
>     [<80538b34>] inet_init+0x184/0x2a0
>     [<80100584>] do_one_initcall+0x174/0x1e0
>     [<8051f348>] kernel_init+0xe4/0x174
>     [<80103d4c>] kernel_thread_helper+0x10/0x18
> unreferenced object 0x8f95e000 (size 4096):
>   comm "swapper", pid 1, jiffies 4294937330 (age 467.240s)
>   hex dump (first 32 bytes):
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>   backtrace:
>     [<80529644>] alloc_large_system_hash+0x2f8/0x410
>     [<8053864c>] udp_table_init+0x4c/0x158
>     [<8053881c>] udplite4_register+0x24/0xa8
>     [<80538b3c>] inet_init+0x18c/0x2a0
>     [<80100584>] do_one_initcall+0x174/0x1e0
>     [<8051f348>] kernel_init+0xe4/0x174
>     [<80103d4c>] kernel_thread_helper+0x10/0x18

These are probably false positives. Since the pointer referring this
block (udp_table) is __read_mostly, is it possible that the
corresponding section gets placed outside the _sdata.._edata range?

diff --git a/arch/mips/include/asm/cache.h b/arch/mips/include/asm/cache.h
index 650ac9b..b4db69f 100644
--- a/arch/mips/include/asm/cache.h
+++ b/arch/mips/include/asm/cache.h
@@ -17,6 +17,6 @@
 #define SMP_CACHE_SHIFT		L1_CACHE_SHIFT
 #define SMP_CACHE_BYTES		L1_CACHE_BYTES
 
-#define __read_mostly __attribute__((__section__(".data.read_mostly")))
+#define __read_mostly __attribute__((__section__(".data..read_mostly")))
 
 #endif /* _ASM_CACHE_H */
diff --git a/arch/mips/kernel/vmlinux.lds.S b/arch/mips/kernel/vmlinux.lds.S
index 570607b..6f6d5d0 100644
--- a/arch/mips/kernel/vmlinux.lds.S
+++ b/arch/mips/kernel/vmlinux.lds.S
@@ -74,6 +74,7 @@ SECTIONS
 		INIT_TASK_DATA(PAGE_SIZE)
 		NOSAVE_DATA
 		CACHELINE_ALIGNED_DATA(1 << CONFIG_MIPS_L1_CACHE_SHIFT)
+		READ_MOSTLY_DATA(1 << CONFIG_MIPS_L1_CACHE_SHIFT)
 		DATA_DATA
 		CONSTRUCTORS
 	}

> unreferenced object 0x8f072000 (size 4096):
>   comm "swapper", pid 1, jiffies 4294937680 (age 463.840s)
>   hex dump (first 32 bytes):
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>   backtrace:
>     [<801ba3d8>] __kmalloc+0x130/0x180
>     [<805461bc>] flow_cache_cpu_prepare+0x50/0xa8
>     [<8053746c>] flow_cache_init_global+0x90/0x138
>     [<80100584>] do_one_initcall+0x174/0x1e0
>     [<8051f348>] kernel_init+0xe4/0x174
>     [<80103d4c>] kernel_thread_helper+0x10/0x18

Same here, flow_cachep pointer is __read_mostly.

-- 
Catalin


--
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/