Re: [PATCH v2 1/7] dmaengine: idxd: fix memory leak in error handling path of idxd_setup_wqs

[Shuai Xue <xueshuai@linux.alibaba.com>]

在 2025/2/19 00:32, Fenghua Yu 写道:
> Hi, Shuai,
> 
> On 2/14/25 21:44, Shuai Xue wrote:
>> Memory allocated for wqs is not freed if an error occurs during
>> idxd_setup_wqs(). To fix it, free the allocated memory in the reverse
>> order of allocation before exiting the function in case of an error.
>>
>> Fixes: a8563a33a5e2 ("dmanegine: idxd: reformat opcap output to match bitmap_parse() input")
>> Signed-off-by: Shuai Xue <xueshuai@linux.alibaba.com>
>> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
>> ---
>>   drivers/dma/idxd/init.c | 20 +++++++++++++++++---
>>   1 file changed, 17 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/dma/idxd/init.c b/drivers/dma/idxd/init.c
>> index b946f78f85e1..b85736fd25bd 100644
>> --- a/drivers/dma/idxd/init.c
>> +++ b/drivers/dma/idxd/init.c
>> @@ -169,8 +169,8 @@ static int idxd_setup_wqs(struct idxd_device *idxd)
>>       idxd->wq_enable_map = bitmap_zalloc_node(idxd->max_wqs, GFP_KERNEL, dev_to_node(dev));
>>       if (!idxd->wq_enable_map) {
>> -        kfree(idxd->wqs);
>> -        return -ENOMEM;
>> +        rc = -ENOMEM;
>> +        goto err_bitmap;
>>       }
>>       for (i = 0; i < idxd->max_wqs; i++) {
>> @@ -191,6 +191,7 @@ static int idxd_setup_wqs(struct idxd_device *idxd)
>>           rc = dev_set_name(conf_dev, "wq%d.%d", idxd->id, wq->id);
>>           if (rc < 0) {
>>               put_device(conf_dev);
>> +            kfree(wq);
>>               goto err;
>>           }
>> @@ -204,6 +205,7 @@ static int idxd_setup_wqs(struct idxd_device *idxd)
>>           wq->wqcfg = kzalloc_node(idxd->wqcfg_size, GFP_KERNEL, dev_to_node(dev));
>>           if (!wq->wqcfg) {
>>               put_device(conf_dev);
>> +            kfree(wq);
>>               rc = -ENOMEM;
>>               goto err;
>>           }
>> @@ -211,7 +213,9 @@ static int idxd_setup_wqs(struct idxd_device *idxd)
>>           if (idxd->hw.wq_cap.op_config) {
>>               wq->opcap_bmap = bitmap_zalloc(IDXD_MAX_OPCAP_BITS, GFP_KERNEL);
>>               if (!wq->opcap_bmap) {
>> +                kfree(wq->wqcfg);
>>                   put_device(conf_dev);
>> +                kfree(wq);
>>                   rc = -ENOMEM;
>>                   goto err;
>>               }
>> @@ -225,11 +229,21 @@ static int idxd_setup_wqs(struct idxd_device *idxd)
>>       return 0;
>>    err:
>> -    while (--i >= 0) {
>> +    while (i-- > 0) {
> 
> Why changed to "i-- > 0" here? Before coming to here, the mem areas allocated for wqs[i] are freed already and there is not need to free them again here, right? 

Yes.

> And if i>1, mem areas for wqs[0] won't be freed and will leak, right?

No, the two ways of writing are equivalent.

#include <stdio.h>

int main()
{
     int i = 1;
     while (i-- > 0)
         printf("freeing i %d\n", i);

     return 0;
}

// console output
// freeing i 0

I will drop this line to avoid confusion.

Thanks.
Shuai