linux-kernel - Re: using ioremap() memory in dma_map

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAHp75Vc0HC9_12GzSN7C0xxD5BYZ+Zi=63oecMotPv_OHORSXw@mail.gmail.com>
Date:   Thu, 22 Nov 2018 13:28:31 +0200
From:   Andy Shevchenko <andy.shevchenko@...il.com>
To:     okitain@...ees.com, Sumit Semwal <sumit.semwal@...aro.org>,
        Christoph Hellwig <hch@....de>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: using ioremap() memory in dma_map_sg()

On Mon, Nov 19, 2018 at 6:31 PM Olga Kitaina <okitain@...ees.com> wrote:
>
> Hi.
> As part of our driver, we want to have some device memory to be
> available to other devices via the DMA-BUF subsystem. The device memory
> is managed by genpool, and mapped from the sram driver via ioremap_wc().
> We initially used sg_set_buf() with the virtual address returned from
> gen_pool_dma_alloc(), but it seems like the wrong thing to do, since the
> virt_addr_valid() check fails. Attempting to use vmalloc_to_page()
> results in a failure down the line, such as:
>
> Unable to handle kernel paging request at virtual address ba405000
> Backtrace:
> (dma_cache_maint_page) from (__dma_page_dev_to_cpu+0x88/0x108)
> (__dma_page_dev_to_cpu) from (arm_dma_map_page+0x5c/0x60)
> (arm_dma_unmap_page) from (arm_dma_map_sg+0x64/0x8c)
> (arm_dma_unmap_sg) from(foo_dmabuf_map+0x60/0x8c [foo])
>
> Is there an appropriate way to put an ioremapped buffer into an SG table?

Olga, I'm not sure you really want DMA-BUF subsystem for your task.
Anyway, I Cc'ed to couple of maintainers, hope they can help you.

-- 
With Best Regards,
Andy Shevchenko