[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <98fdfb0c-0207-9a0a-4875-b834cfd1b2c3@oracle.com>
Date: Fri, 4 Jan 2019 09:34:33 +0800
From: Dongli Zhang <dongli.zhang@...cle.com>
To: iommu@...ts.linux-foundation.org, linux-kernel@...r.kernel.org,
konrad.wilk@...cle.com
Cc: Joe Jin <joe.jin@...cle.com>,
Tim Chen <tim.c.chen@...ux.intel.com>, hch@....de,
m.szyprowski@...sung.com, robin.murphy@....com
Subject: Re: [PATCH v2 1/2] swiotlb: add debugfs to track swiotlb buffer usage
Hi Konrad,
Would you please take a look on those two patches?
In addition, there is another patch correcting an error in comment.
https://lkml.org/lkml/2018/12/5/1721
Thank you very much!
Dongli Zhang
On 12/11/2018 05:05 AM, Joe Jin wrote:
> On 12/10/18 12:00 PM, Tim Chen wrote:
>>> @@ -528,6 +538,9 @@ phys_addr_t swiotlb_tbl_map_single(struct device *hwdev,
>>> dev_warn(hwdev, "swiotlb buffer is full (sz: %zd bytes)\n", size);
>>> return SWIOTLB_MAP_ERROR;
>>> found:
>>> +#ifdef CONFIG_DEBUG_FS
>>> + io_tlb_used += nslots;
>>> +#endif
>> One nit I have about this patch is there are too many CONFIG_DEBUG_FS.
>>
>> For example here, instead of io_tlb_used, we can have a macro defined,
>> perhaps something like inc_iotlb_used(nslots). It can be placed in the
>> same section that swiotlb_create_debugfs is defined so there's a single
>> place where all the CONFIG_DEBUG_FS stuff is located.
>>
>> Then define inc_iotlb_used to be null when we don't have
>> CONFIG_DEBUG_FS.
>>
>
> Dongli had removed above ifdef/endif on his next patch, "[PATCH v2 2/2]
> swiotlb: checking whether swiotlb buffer is full with io_tlb_used"
>
> Thanks,
> Joe
>
Powered by blists - more mailing lists