Hi,

On Mon, 7 Feb 2011, Vishwanath Sripathy wrote:
Temporary strings with volt_* file names should be released after the
debugfs entries are created. While at it, also simplify the string
allocation, and use just snprintf() to create the name.

The patch eliminates kmemleak reports with the following stack trace
(multiple objects depending on HW):

unreferenced object 0xcedbc5a0 (size 64):
  comm "swapper", pid 1, jiffies 4294929375 (age 423.734s)
  hex dump (first 32 bytes):
    76 6f 6c 74 5f 39 37 35 30 30 30 00 00 00 00 00  volt_975000.....
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<c012fee0>] create_object+0x104/0x208
    [<c012dbc8>] kmem_cache_alloc_trace+0xf0/0x17c
    [<c0013f64>] omap_sr_probe+0x314/0x420
    [<c02a1724>] platform_drv_probe+0x18/0x1c
    [<c02a088c>] driver_probe_device+0xc8/0x188
    [<c02a09b4>] __driver_attach+0x68/0x8c
    [<c02a00ac>] bus_for_each_dev+0x44/0x74
    [<c029f9e0>] bus_add_driver+0xa0/0x228
    [<c02a0cac>] driver_register+0xa8/0x130
    [<c02a1b2c>] platform_driver_probe+0x18/0x8c
    [<c0013c1c>] sr_init+0x40/0x74
    [<c005a554>] do_one_initcall+0xc8/0x1a0
    [<c00084f4>] kernel_init+0x150/0x218
    [<c0065d64>] kernel_thread_exit+0x0/0x8
    [<ffffffff>] 0xffffffff

Signed-off-by: Aaro Koskinen <aaro.koski...@nokia.com>
---

v2: Get rid of kmalloc().

 arch/arm/mach-omap2/smartreflex.c |   15 +++------------
 1 files changed, 3 insertions(+), 12 deletions(-)

diff --git a/arch/arm/mach-omap2/smartreflex.c b/arch/arm/mach-
omap2/smartreflex.c
index 77ecebf..e54db84 100644
--- a/arch/arm/mach-omap2/smartreflex.c
+++ b/arch/arm/mach-omap2/smartreflex.c
@@ -927,19 +927,10 @@ static int __init omap_sr_probe(struct
platform_device *pdev)
        }

        for (i = 0; i < sr_info->nvalue_count; i++) {
-               char *name;
-               char volt_name[32];
-
-               name = kzalloc(NVALUE_NAME_LEN + 1, GFP_KERNEL);
-               if (!name) {
-                       dev_err(&pdev->dev, "%s: Unable to allocate
memory"
-                               " for n-value directory name\n",
__func__);
-                       return -ENOMEM;
-               }
+               char name[NVALUE_NAME_LEN + 1];
Is there any need to keep this inside the loop?

That's the correct place to declare it - the variable's scope should be
limited to the block where it's used.

Defining it in the beginning of function might save a few cpu cycles.

I don't think so.

A.
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to