[PATCH v3] platform/x86: intel_pmc_ipc: fix ACPI buffer memory leak

yongxin.liu@windriver.com posted 1 patch 3 days, 17 hours ago
There is a newer version of this series
include/linux/platform_data/x86/intel_pmc_ipc.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
[PATCH v3] platform/x86: intel_pmc_ipc: fix ACPI buffer memory leak
Posted by yongxin.liu@windriver.com 3 days, 17 hours ago
From: Yongxin Liu <yongxin.liu@windriver.com>

The intel_pmc_ipc() function uses ACPI_ALLOCATE_BUFFER to allocate memory
for the ACPI evaluation result but never frees it, causing a 192-byte
memory leak on each call.

This leak is triggered during network interface initialization when the
stmmac driver calls intel_mac_finish() -> intel_pmc_ipc().

  unreferenced object 0xffff96a848d6ea80 (size 192):
    comm "dhcpcd", pid 541, jiffies 4294684345
    hex dump (first 32 bytes):
      04 00 00 00 05 00 00 00 98 ea d6 48 a8 96 ff ff  ...........H....
      00 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
    backtrace (crc b1564374):
      kmemleak_alloc+0x2d/0x40
      __kmalloc_noprof+0x2fa/0x730
      acpi_ut_initialize_buffer+0x83/0xc0
      acpi_evaluate_object+0x29a/0x2f0
      intel_pmc_ipc+0xfd/0x170
      intel_mac_finish+0x168/0x230
      stmmac_mac_finish+0x3d/0x50
      phylink_major_config+0x22b/0x5b0
      phylink_mac_initial_config.constprop.0+0xf1/0x1b0
      phylink_start+0x8e/0x210
      __stmmac_open+0x12c/0x2b0
      stmmac_open+0x23c/0x380
      __dev_open+0x11d/0x2c0
      __dev_change_flags+0x1d2/0x250
      netif_change_flags+0x2b/0x70
      dev_change_flags+0x40/0xb0

Add __free(kfree) for ACPI object to properly release the allocated buffer.

Cc: stable@vger.kernel.org
Fixes: 7e2f7e25f6ff ("arch: x86: add IPC mailbox accessor function and add SoC register access")
Signed-off-by: Yongxin Liu <yongxin.liu@windriver.com>
---
V2->V3:
Use __free(kfree) instead of goto and kfree();

V1->V2:
Cover all potential paths for kfree();
---
 include/linux/platform_data/x86/intel_pmc_ipc.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/include/linux/platform_data/x86/intel_pmc_ipc.h b/include/linux/platform_data/x86/intel_pmc_ipc.h
index 1d34435b7001..cf0b78048b0e 100644
--- a/include/linux/platform_data/x86/intel_pmc_ipc.h
+++ b/include/linux/platform_data/x86/intel_pmc_ipc.h
@@ -9,6 +9,7 @@
 #ifndef INTEL_PMC_IPC_H
 #define INTEL_PMC_IPC_H
 #include <linux/acpi.h>
+#include <linux/cleanup.h>
 
 #define IPC_SOC_REGISTER_ACCESS			0xAA
 #define IPC_SOC_SUB_CMD_READ			0x00
@@ -48,7 +49,7 @@ static inline int intel_pmc_ipc(struct pmc_ipc_cmd *ipc_cmd, struct pmc_ipc_rbuf
 		{.type = ACPI_TYPE_INTEGER,},
 	};
 	struct acpi_object_list arg_list = { PMC_IPCS_PARAM_COUNT, params };
-	union acpi_object *obj;
+	union acpi_object *obj __free(kfree) = NULL;
 	int status;
 
 	if (!ipc_cmd || !rbuf)
-- 
2.46.2
Re: [PATCH v3] platform/x86: intel_pmc_ipc: fix ACPI buffer memory leak
Posted by Krzysztof Kozlowski 2 days, 6 hours ago
On 28/11/2025 04:32, yongxin.liu@windriver.com wrote:
> From: Yongxin Liu <yongxin.liu@windriver.com>
> 
> The intel_pmc_ipc() function uses ACPI_ALLOCATE_BUFFER to allocate memory
> for the ACPI evaluation result but never frees it, causing a 192-byte
> memory leak on each call.
> 
> This leak is triggered during network interface initialization when the
> stmmac driver calls intel_mac_finish() -> intel_pmc_ipc().
> 
>   unreferenced object 0xffff96a848d6ea80 (size 192):
>     comm "dhcpcd", pid 541, jiffies 4294684345
>     hex dump (first 32 bytes):
>       04 00 00 00 05 00 00 00 98 ea d6 48 a8 96 ff ff  ...........H....
>       00 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
>     backtrace (crc b1564374):
>       kmemleak_alloc+0x2d/0x40
>       __kmalloc_noprof+0x2fa/0x730
>       acpi_ut_initialize_buffer+0x83/0xc0
>       acpi_evaluate_object+0x29a/0x2f0
>       intel_pmc_ipc+0xfd/0x170
>       intel_mac_finish+0x168/0x230
>       stmmac_mac_finish+0x3d/0x50
>       phylink_major_config+0x22b/0x5b0
>       phylink_mac_initial_config.constprop.0+0xf1/0x1b0
>       phylink_start+0x8e/0x210
>       __stmmac_open+0x12c/0x2b0
>       stmmac_open+0x23c/0x380
>       __dev_open+0x11d/0x2c0
>       __dev_change_flags+0x1d2/0x250
>       netif_change_flags+0x2b/0x70
>       dev_change_flags+0x40/0xb0
> 
> Add __free(kfree) for ACPI object to properly release the allocated buffer.
> 
> Cc: stable@vger.kernel.org
> Fixes: 7e2f7e25f6ff ("arch: x86: add IPC mailbox accessor function and add SoC register access")
> Signed-off-by: Yongxin Liu <yongxin.liu@windriver.com>
> ---
> V2->V3:
> Use __free(kfree) instead of goto and kfree();
> 
> V1->V2:
> Cover all potential paths for kfree();
> ---
>  include/linux/platform_data/x86/intel_pmc_ipc.h | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/platform_data/x86/intel_pmc_ipc.h b/include/linux/platform_data/x86/intel_pmc_ipc.h
> index 1d34435b7001..cf0b78048b0e 100644
> --- a/include/linux/platform_data/x86/intel_pmc_ipc.h
> +++ b/include/linux/platform_data/x86/intel_pmc_ipc.h
> @@ -9,6 +9,7 @@
>  #ifndef INTEL_PMC_IPC_H
>  #define INTEL_PMC_IPC_H
>  #include <linux/acpi.h>
> +#include <linux/cleanup.h>
>  
>  #define IPC_SOC_REGISTER_ACCESS			0xAA
>  #define IPC_SOC_SUB_CMD_READ			0x00
> @@ -48,7 +49,7 @@ static inline int intel_pmc_ipc(struct pmc_ipc_cmd *ipc_cmd, struct pmc_ipc_rbuf
>  		{.type = ACPI_TYPE_INTEGER,},
>  	};
>  	struct acpi_object_list arg_list = { PMC_IPCS_PARAM_COUNT, params };
> -	union acpi_object *obj;
> +	union acpi_object *obj __free(kfree) = NULL;


This is undesired syntax explicitly documented as one to avoid. Please
don't use cleanup.h if you do not intend to follow it because it does
not make the code simpler. The rule of explicit (useful, not NULL)
constructor

Best regards,
Krzysztof
Re: [PATCH v3] platform/x86: intel_pmc_ipc: fix ACPI buffer memory leak
Posted by Ilpo Järvinen 3 days, 11 hours ago
On Fri, 28 Nov 2025, yongxin.liu@windriver.com wrote:

> From: Yongxin Liu <yongxin.liu@windriver.com>
> 
> The intel_pmc_ipc() function uses ACPI_ALLOCATE_BUFFER to allocate memory
> for the ACPI evaluation result but never frees it, causing a 192-byte
> memory leak on each call.
> 
> This leak is triggered during network interface initialization when the
> stmmac driver calls intel_mac_finish() -> intel_pmc_ipc().
> 
>   unreferenced object 0xffff96a848d6ea80 (size 192):
>     comm "dhcpcd", pid 541, jiffies 4294684345
>     hex dump (first 32 bytes):
>       04 00 00 00 05 00 00 00 98 ea d6 48 a8 96 ff ff  ...........H....
>       00 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
>     backtrace (crc b1564374):
>       kmemleak_alloc+0x2d/0x40
>       __kmalloc_noprof+0x2fa/0x730
>       acpi_ut_initialize_buffer+0x83/0xc0
>       acpi_evaluate_object+0x29a/0x2f0
>       intel_pmc_ipc+0xfd/0x170
>       intel_mac_finish+0x168/0x230
>       stmmac_mac_finish+0x3d/0x50
>       phylink_major_config+0x22b/0x5b0
>       phylink_mac_initial_config.constprop.0+0xf1/0x1b0
>       phylink_start+0x8e/0x210
>       __stmmac_open+0x12c/0x2b0
>       stmmac_open+0x23c/0x380
>       __dev_open+0x11d/0x2c0
>       __dev_change_flags+0x1d2/0x250
>       netif_change_flags+0x2b/0x70
>       dev_change_flags+0x40/0xb0
> 
> Add __free(kfree) for ACPI object to properly release the allocated buffer.
> 
> Cc: stable@vger.kernel.org
> Fixes: 7e2f7e25f6ff ("arch: x86: add IPC mailbox accessor function and add SoC register access")
> Signed-off-by: Yongxin Liu <yongxin.liu@windriver.com>
> ---
> V2->V3:
> Use __free(kfree) instead of goto and kfree();
> 
> V1->V2:
> Cover all potential paths for kfree();
> ---
>  include/linux/platform_data/x86/intel_pmc_ipc.h | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/platform_data/x86/intel_pmc_ipc.h b/include/linux/platform_data/x86/intel_pmc_ipc.h
> index 1d34435b7001..cf0b78048b0e 100644
> --- a/include/linux/platform_data/x86/intel_pmc_ipc.h
> +++ b/include/linux/platform_data/x86/intel_pmc_ipc.h
> @@ -9,6 +9,7 @@
>  #ifndef INTEL_PMC_IPC_H
>  #define INTEL_PMC_IPC_H
>  #include <linux/acpi.h>
> +#include <linux/cleanup.h>
>  
>  #define IPC_SOC_REGISTER_ACCESS			0xAA
>  #define IPC_SOC_SUB_CMD_READ			0x00
> @@ -48,7 +49,7 @@ static inline int intel_pmc_ipc(struct pmc_ipc_cmd *ipc_cmd, struct pmc_ipc_rbuf
>  		{.type = ACPI_TYPE_INTEGER,},
>  	};
>  	struct acpi_object_list arg_list = { PMC_IPCS_PARAM_COUNT, params };
> -	union acpi_object *obj;
> +	union acpi_object *obj __free(kfree) = NULL;

Please declare it where the value is getting assigned to it like I 
instructed in v1. While not strictly necessary here, I want us to 
reinforce the only correct pattern to use cleanup.h helpers at every
usage site.

The placement matters when there is more than once cleanup.h thing done 
within a function. The cleanup order depends on the order you declared the 
variables.

>  	int status;
>  
>  	if (!ipc_cmd || !rbuf)
> 

-- 
 i.