A different approach tried to fix the memory leak in the
regmap_debugfs_init() function about one year ago. But it added the missing
kfree() at the wrong code path. Even with this approach applied,
'kmemleak' still finds this memory leak (at least on my i.MX6Q SoC):
unreferenced object 0xec548c80 (size 64):
comm "swapper/0", pid 1, jiffies 4294937356 (age 3385.690s)
hex dump (first 32 bytes):
64 75 6d 6d 79 2d 69 6f 6d 75 78 63 2d 67 70 72 dummy-iomuxc-gpr
40 32 30 65 30 30 30 30 00 7d cc fa 7c df bf 7d @20e0000.}..|..}
backtrace:
[<703904e1>] __kmalloc_track_caller+0x1dc/0x43c
[<523053d0>] kvasprintf+0x64/0xd8
[<0d94bbe5>] kasprintf+0x38/0x5c
[<4ba65bc1>] regmap_debugfs_init+0xa8/0x330
[<dfcf84e9>] __regmap_init+0x890/0xd84
[<750d0959>] __regmap_init_mmio_clk+0x50/0x5c
[<02a69967>] of_syscon_register+0x194/0x2a8
[<8d9f337a>] device_node_get_regmap+0x9c/0xa4
[<97914327>] syscon_node_to_regmap+0x40/0x44
[<b1677f4e>] syscon_regmap_lookup_by_compatible+0x30/0x3c
[<0b2d4178>] imx6_pm_common_init.constprop.0+0x4c/0x7c
[<76253dc8>] imx6q_pm_init+0x18/0x1c
[<1d2cec28>] imx6q_init_machine+0x120/0x2bc
[<7386bff5>] customize_machine+0x30/0x38
[<bc13c21d>] do_one_initcall+0x68/0x2f8
[<a31f6aea>] kernel_init_freeable+0x24c/0x2f0
For this specific i.MX device the regmap_debugfs_init() function is called
twice for the same object:
- first call: with name = "iomuxc-gpr@20e0000", but the underlying device
has no name yet. So, regmap_debugfs_init() allocates the string
"dummy-iomuxc-gpr@20e0000" into 'map->debugfs_name' (and this string is
the one 'kmemleak' reports in its output).
- second call: with name = "gpr" and the underlying device now has the
name "20e0000.iomuxc". So, regmap_debugfs_init() again allocates
a string "20e0000.iomuxc-gpr" into 'map->debugfs_name'.
Since the underlying regmap structure is allocated via kzalloc() the
'map->debugfs_name' member should be NULL on the first call. So, it should
be safe to kfree() this member unconditionally prior allocating a new
string for this member.
Fixes: 2899872b627e ("regmap: debugfs: Fix memory leak in regmap_debugfs_init")
Signed-off-by: Juergen Borleis <[email protected]>
---
drivers/base/regmap/regmap-debugfs.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/base/regmap/regmap-debugfs.c b/drivers/base/regmap/regmap-debugfs.c
index 089e5dc7144a..81c8fff8c09f 100644
--- a/drivers/base/regmap/regmap-debugfs.c
+++ b/drivers/base/regmap/regmap-debugfs.c
@@ -577,6 +577,7 @@ void regmap_debugfs_init(struct regmap *map, const char *name)
devname = dev_name(map->dev);
if (name) {
+ kfree(map->debugfs_name);
map->debugfs_name = kasprintf(GFP_KERNEL, "%s-%s",
devname, name);
name = map->debugfs_name;
--
2.20.1
On Tue, Jun 30, 2020 at 12:06:50PM +0200, Juergen Borleis wrote:
> unreferenced object 0xec548c80 (size 64):
> comm "swapper/0", pid 1, jiffies 4294937356 (age 3385.690s)
> hex dump (first 32 bytes):
> 64 75 6d 6d 79 2d 69 6f 6d 75 78 63 2d 67 70 72 dummy-iomuxc-gpr
> 40 32 30 65 30 30 30 30 00 7d cc fa 7c df bf 7d @20e0000.}..|..}
> backtrace:
> [<703904e1>] __kmalloc_track_caller+0x1dc/0x43c
> [<523053d0>] kvasprintf+0x64/0xd8
> [<0d94bbe5>] kasprintf+0x38/0x5c
Please think hard before including complete backtraces in upstream
reports, they are very large and contain almost no useful information
relative to their size so often obscure the relevant content in your
message. If part of the backtrace is usefully illustrative (it often is
for search engines if nothing else) then it's usually better to pull out
the relevant sections.
> For this specific i.MX device the regmap_debugfs_init() function is called
> twice for the same object:
> - first call: with name = "iomuxc-gpr@20e0000", but the underlying device
> has no name yet. So, regmap_debugfs_init() allocates the string
> "dummy-iomuxc-gpr@20e0000" into 'map->debugfs_name' (and this string is
> the one 'kmemleak' reports in its output).
> - second call: with name = "gpr" and the underlying device now has the
> name "20e0000.iomuxc". So, regmap_debugfs_init() again allocates
> a string "20e0000.iomuxc-gpr" into 'map->debugfs_name'.
This sounds like an issue in and of itself - why are we initializing the
same regmap twice without first destroying it?