[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZNUK8nWnUYB6B4Kg@infradead.org>
Date: Thu, 10 Aug 2023 09:06:10 -0700
From: Christoph Hellwig <hch@...radead.org>
To: Richard Weinberger <richard@....at>
Cc: linux-mtd@...ts.infradead.org,
Christoph Hellwig <hch@...radead.org>,
Stephan Wurm <stephan.wurm@...berle.de>,
stable@...r.kernel.org, Miquel Raynal <miquel.raynal@...tlin.com>,
Vignesh Raghavendra <vigneshr@...com>,
Oliver Neukum <oliver@...kum.org>,
Ali Akcaagac <aliakc@....de>,
Jamie Lenehan <lenehan@...bble.org>,
"James E.J. Bottomley" <jejb@...ux.ibm.com>,
"Martin K. Petersen" <martin.petersen@...cle.com>,
Ezequiel Garcia <ezequiel@...guardiasur.com.ar>,
linux-kernel@...r.kernel.org, linux-scsi@...r.kernel.org
Subject: Re: [PATCH 1/7] ubi: block: Refactor sg list processing for highmem
On Thu, Aug 10, 2023 at 06:00:12PM +0200, Richard Weinberger wrote:
> Currently sg_virt() is used while filling the sg list from LEB data.
> This approach cannot work with highmem.
>
> Refactor ubi_eba_read_leb_sg() to use kmap_atomic() for sg list
> access.
> Since kmap_atomic() disables preempt a bounce buffer is needed.
> kmap_local_page() is not used to allow easy backporting of this patch
> to older kernels.
>
> The followup patches in this series will switch to kmap_sg()
> and we can remove our own helper and the bounce buffer.
Please just use kmap_local and avoid the bounce buffering.
Powered by blists - more mailing lists