From: Diana Madalina Craciun <diana.craciun@nxp.com>
To: Leo Li <leoyang.li@nxp.com>, Ioana Ciornei <ioana.ciornei@nxp.com>
Cc: Youri Querry <youri.querry_1@nxp.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
Arnd Bergmann <arnd@arndb.de>
Subject: Re: [PATCH 4/5] soc: fsl: dpio: fix qbman alignment error in the virtualization context
Date: Thu, 28 Oct 2021 14:32:09 +0300 [thread overview]
Message-ID: <69bd66c8-d4f5-7649-33bc-71a507c65ce1@nxp.com> (raw)
In-Reply-To: <AS8PR04MB8946734C740B202DC3CA88F08F849@AS8PR04MB8946.eurprd04.prod.outlook.com>
Hi Leo,
Just drop this patch. I realize that the implementation is not correct.
The problem should be resolved (if possible) elsewhere.
I will open another thread on the relevant lists regarding this problem.
Diana
On 10/27/2021 2:47 AM, Leo Li wrote:
>
>> -----Original Message-----
>> From: Ioana Ciornei <ioana.ciornei@nxp.com>
>> Sent: Monday, October 18, 2021 10:11 AM
>> To: Leo Li <leoyang.li@nxp.com>
>> Cc: Youri Querry <youri.querry_1@nxp.com>; linux-kernel@vger.kernel.org;
>> Diana Madalina Craciun <diana.craciun@nxp.com>; Ioana Ciornei
>> <ioana.ciornei@nxp.com>
>> Subject: [PATCH 4/5] soc: fsl: dpio: fix qbman alignment error in the
>> virtualization context
>>
>> From: Diana Craciun <diana.craciun@nxp.com>
>>
>> When running as a guest, under KVM, the CENA region is mapped as device
>> memory, so uncacheable. When the memory is mapped as device memory,
>> the unaligned accesses are not allowed. Memcpy is optimized to transfer 8
>> bytes at a time regardless of the start address and might cause alignment
>> issues.
>>
>> Signed-off-by: Diana Craciun <diana.craciun@nxp.com>
>> Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com>
> We get the following feedback from Arnd about this patch. Could you respond to the comments?
>
> "This patch looks very suspicious to me, I don't think it's generally safe to
> use memcpy_toio() on a normal pointer, as the __iomem tokens may
> be in a separate address range, even though this currently works
> on arm64. Adding the (__iomem void *) cast without a comment that
> explains why it's added seems similarly wrong, and finally the
> changeset text does not seem to match what the code does:
>
> According to the text, the pointer is to a virtual address mapped as
> "device memory" (i.e. PROT_DEVICE_nGnRE or PROT_DEVICE_nGnRnE),
> but the code suggests it's actually write-combining normal
> (PROT_NORMAL_NC)."
>
>> ---
>> drivers/soc/fsl/dpio/qbman-portal.c | 24 ++++++++++++------------
>> 1 file changed, 12 insertions(+), 12 deletions(-)
>>
>> diff --git a/drivers/soc/fsl/dpio/qbman-portal.c
>> b/drivers/soc/fsl/dpio/qbman-portal.c
>> index 3fd54611ed98..ef9cafd12534 100644
>> --- a/drivers/soc/fsl/dpio/qbman-portal.c
>> +++ b/drivers/soc/fsl/dpio/qbman-portal.c
>> @@ -679,9 +679,9 @@ int qbman_swp_enqueue_multiple_direct(struct
>> qbman_swp *s,
>> for (i = 0; i < num_enqueued; i++) {
>> p = (s->addr_cena + QBMAN_CENA_SWP_EQCR(eqcr_pi &
>> half_mask));
>> /* Skip copying the verb */
>> - memcpy(&p[1], &cl[1], EQ_DESC_SIZE_WITHOUT_FD - 1);
>> - memcpy(&p[EQ_DESC_SIZE_FD_START/sizeof(uint32_t)],
>> - &fd[i], sizeof(*fd));
>> + memcpy_toio((__iomem void *)&p[1], &cl[1],
>> EQ_DESC_SIZE_WITHOUT_FD - 1);
>> + memcpy_toio((__iomem void
>> *)&p[EQ_DESC_SIZE_FD_START / sizeof(uint32_t)],
>> + &fd[i], sizeof(*fd));
>> eqcr_pi++;
>> }
>>
>> @@ -763,9 +763,9 @@ int
>> qbman_swp_enqueue_multiple_mem_back(struct qbman_swp *s,
>> for (i = 0; i < num_enqueued; i++) {
>> p = (s->addr_cena + QBMAN_CENA_SWP_EQCR(eqcr_pi &
>> half_mask));
>> /* Skip copying the verb */
>> - memcpy(&p[1], &cl[1], EQ_DESC_SIZE_WITHOUT_FD - 1);
>> - memcpy(&p[EQ_DESC_SIZE_FD_START/sizeof(uint32_t)],
>> - &fd[i], sizeof(*fd));
>> + memcpy_toio((__iomem void *)&p[1], &cl[1],
>> EQ_DESC_SIZE_WITHOUT_FD - 1);
>> + memcpy_toio((__iomem void
>> *)&p[EQ_DESC_SIZE_FD_START / sizeof(uint32_t)],
>> + &fd[i], sizeof(*fd));
>> eqcr_pi++;
>> }
>>
>> @@ -837,9 +837,9 @@ int
>> qbman_swp_enqueue_multiple_desc_direct(struct qbman_swp *s,
>> p = (s->addr_cena + QBMAN_CENA_SWP_EQCR(eqcr_pi &
>> half_mask));
>> cl = (uint32_t *)(&d[i]);
>> /* Skip copying the verb */
>> - memcpy(&p[1], &cl[1], EQ_DESC_SIZE_WITHOUT_FD - 1);
>> - memcpy(&p[EQ_DESC_SIZE_FD_START/sizeof(uint32_t)],
>> - &fd[i], sizeof(*fd));
>> + memcpy_toio((__iomem void *)&p[1], &cl[1],
>> EQ_DESC_SIZE_WITHOUT_FD - 1);
>> + memcpy_toio((__iomem void
>> *)&p[EQ_DESC_SIZE_FD_START / sizeof(uint32_t)],
>> + &fd[i], sizeof(*fd));
>> eqcr_pi++;
>> }
>>
>> @@ -907,9 +907,9 @@ int
>> qbman_swp_enqueue_multiple_desc_mem_back(struct qbman_swp *s,
>> p = (s->addr_cena + QBMAN_CENA_SWP_EQCR(eqcr_pi &
>> half_mask));
>> cl = (uint32_t *)(&d[i]);
>> /* Skip copying the verb */
>> - memcpy(&p[1], &cl[1], EQ_DESC_SIZE_WITHOUT_FD - 1);
>> - memcpy(&p[EQ_DESC_SIZE_FD_START/sizeof(uint32_t)],
>> - &fd[i], sizeof(*fd));
>> + memcpy_toio((__iomem void *)&p[1], &cl[1],
>> EQ_DESC_SIZE_WITHOUT_FD - 1);
>> + memcpy_toio((__iomem void
>> *)&p[EQ_DESC_SIZE_FD_START / sizeof(uint32_t)],
>> + &fd[i], sizeof(*fd));
>> eqcr_pi++;
>> }
>>
>> --
>> 2.33.1
next prev parent reply other threads:[~2021-10-28 11:32 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-18 15:10 [PATCH 0/5] soc: fsl: various fixes Ioana Ciornei
2021-10-18 15:10 ` [PATCH 1/5] soc: fsl: dpio: use an explicit NULL instead of 0 Ioana Ciornei
2021-10-18 15:10 ` [PATCH 2/5] soc: fsl: dpio: rename the enqueue descriptor variable Ioana Ciornei
2021-10-18 15:10 ` [PATCH 3/5] soc: fsl: dpio: fix kernel-doc warnings Ioana Ciornei
2021-10-18 15:10 ` [PATCH 4/5] soc: fsl: dpio: fix qbman alignment error in the virtualization context Ioana Ciornei
2021-10-26 23:47 ` Leo Li
2021-10-28 11:32 ` Diana Madalina Craciun [this message]
2021-10-18 15:10 ` [PATCH 5/5] soc: fsl: dpaa2-console: free buffer before returning from dpaa2_console_read Ioana Ciornei
2021-10-21 23:30 ` [PATCH 0/5] soc: fsl: various fixes Leo Li
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=69bd66c8-d4f5-7649-33bc-71a507c65ce1@nxp.com \
--to=diana.craciun@nxp.com \
--cc=arnd@arndb.de \
--cc=ioana.ciornei@nxp.com \
--cc=leoyang.li@nxp.com \
--cc=linux-kernel@vger.kernel.org \
--cc=youri.querry_1@nxp.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®