diff mbox series

ACPI / APEI: release resources if gen_pool_add fails

Message ID 1560505783-130606-1-git-send-email-zhangliguang@linux.alibaba.com
State Superseded, archived
Headers show
Series ACPI / APEI: release resources if gen_pool_add fails | expand

Commit Message

Liguang Zhang June 14, 2019, 9:49 a.m. UTC
To avoid memory leaks, destroy ghes_estatus_pool and release memory
allocated via vmalloc() on errors in ghes_estatus_pool_init().

Signed-off-by: liguang.zlg <zhangliguang@linux.alibaba.com>
---
 drivers/acpi/apei/ghes.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

Comments

James Morse June 14, 2019, 2:51 p.m. UTC | #1
Hi Liguang,

On 14/06/2019 10:49, luanshi wrote:
> To avoid memory leaks, destroy ghes_estatus_pool and release memory
> allocated via vmalloc() on errors in ghes_estatus_pool_init().
> 
> Signed-off-by: liguang.zlg <zhangliguang@linux.alibaba.com>

(I'm surprised your name has a '.' in it!)

Nit: This is v2. Please add a version number in the subject, e.g.:
| [PATCH v2] ACPI / APEI: release resources if gen_pool_add fails

This makes it easy for reviewers to know which is the latest. git format-patch will do
this for you if you add '-v 2' to its command-line.


> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index 993940d..8472c96 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -163,8 +164,10 @@ int ghes_estatus_pool_init(int num_ghes)
>  
>  	ghes_estatus_pool_size_request = PAGE_ALIGN(len);
>  	addr = (unsigned long)vmalloc(PAGE_ALIGN(len));
> -	if (!addr)
> +	if (!addr) {
> +		gen_pool_destroy(ghes_estatus_pool);
>  		return -ENOMEM;
> +	}
>  
>  	/*
>  	 * New allocation must be visible in all pgd before it can be found by
> @@ -172,7 +175,12 @@ int ghes_estatus_pool_init(int num_ghes)
>  	 */
>  	vmalloc_sync_all();
>  
> -	return gen_pool_add(ghes_estatus_pool, addr, PAGE_ALIGN(len), -1);
> +	rc = gen_pool_add(ghes_estatus_pool, addr, PAGE_ALIGN(len), -1);
> +	if (rc) {

> +		vfree(addr);

addr here is unsigned long, but vfree() wants a void *.

vfree() first leaves us with a pool containing memory we've vfree()d, which doesn't feel
like a good state to step through.
Can we vfree() after gen_pool_destroy()?


> +		gen_pool_destroy(ghes_estatus_pool);
> +	}
> +	return rc;
>  }


With that:
Reviewed-by: James Morse <james.morse@arm.com>
Tested-by: James Morse <james.morse@arm.com>


Thanks for cleaning this up!

James
Liguang Zhang June 15, 2019, 12:55 a.m. UTC | #2
在 2019/6/14 22:51, James Morse 写道:
> Hi Liguang,
>
> On 14/06/2019 10:49, luanshi wrote:
>> To avoid memory leaks, destroy ghes_estatus_pool and release memory
>> allocated via vmalloc() on errors in ghes_estatus_pool_init().
>>
>> Signed-off-by: liguang.zlg <zhangliguang@linux.alibaba.com>
> (I'm surprised your name has a '.' in it!)
>
> Nit: This is v2. Please add a version number in the subject, e.g.:
> | [PATCH v2] ACPI / APEI: release resources if gen_pool_add fails
>
> This makes it easy for reviewers to know which is the latest. git format-patch will do
> this for you if you add '-v 2' to its command-line.
>
>
>> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
>> index 993940d..8472c96 100644
>> --- a/drivers/acpi/apei/ghes.c
>> +++ b/drivers/acpi/apei/ghes.c
>> @@ -163,8 +164,10 @@ int ghes_estatus_pool_init(int num_ghes)
>>   
>>   	ghes_estatus_pool_size_request = PAGE_ALIGN(len);
>>   	addr = (unsigned long)vmalloc(PAGE_ALIGN(len));
>> -	if (!addr)
>> +	if (!addr) {
>> +		gen_pool_destroy(ghes_estatus_pool);
>>   		return -ENOMEM;
>> +	}
>>   
>>   	/*
>>   	 * New allocation must be visible in all pgd before it can be found by
>> @@ -172,7 +175,12 @@ int ghes_estatus_pool_init(int num_ghes)
>>   	 */
>>   	vmalloc_sync_all();
>>   
>> -	return gen_pool_add(ghes_estatus_pool, addr, PAGE_ALIGN(len), -1);
>> +	rc = gen_pool_add(ghes_estatus_pool, addr, PAGE_ALIGN(len), -1);
>> +	if (rc) {
>> +		vfree(addr);
> addr here is unsigned long, but vfree() wants a void *.
>
> vfree() first leaves us with a pool containing memory we've vfree()d, which doesn't feel
> like a good state to step through.
> Can we vfree() after gen_pool_destroy()?

Hi James,

  I will modify this patch by your suggestion and send the patch v3 for 
review.

Thanks,

Liguang


>
>
>> +		gen_pool_destroy(ghes_estatus_pool);
>> +	}
>> +	return rc;
>>   }
>
> With that:
> Reviewed-by: James Morse <james.morse@arm.com>
> Tested-by: James Morse <james.morse@arm.com>
>
>
> Thanks for cleaning this up!
>
> James
diff mbox series

Patch

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 993940d..8472c96 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -153,6 +153,7 @@  static void ghes_unmap(void __iomem *vaddr, enum fixed_addresses fixmap_idx)
 int ghes_estatus_pool_init(int num_ghes)
 {
 	unsigned long addr, len;
+	int rc = 0;
 
 	ghes_estatus_pool = gen_pool_create(GHES_ESTATUS_POOL_MIN_ALLOC_ORDER, -1);
 	if (!ghes_estatus_pool)
@@ -163,8 +164,10 @@  int ghes_estatus_pool_init(int num_ghes)
 
 	ghes_estatus_pool_size_request = PAGE_ALIGN(len);
 	addr = (unsigned long)vmalloc(PAGE_ALIGN(len));
-	if (!addr)
+	if (!addr) {
+		gen_pool_destroy(ghes_estatus_pool);
 		return -ENOMEM;
+	}
 
 	/*
 	 * New allocation must be visible in all pgd before it can be found by
@@ -172,7 +175,12 @@  int ghes_estatus_pool_init(int num_ghes)
 	 */
 	vmalloc_sync_all();
 
-	return gen_pool_add(ghes_estatus_pool, addr, PAGE_ALIGN(len), -1);
+	rc = gen_pool_add(ghes_estatus_pool, addr, PAGE_ALIGN(len), -1);
+	if (rc) {
+		vfree(addr);
+		gen_pool_destroy(ghes_estatus_pool);
+	}
+	return rc;
 }
 
 static int map_gen_v2(struct ghes *ghes)