From: michael.christie@oracle.com
To: John Garry <john.g.garry@oracle.com>,
Bart Van Assche <bvanassche@acm.org>,
Juergen Gross <jgross@suse.com>,
"Martin K. Petersen" <martin.petersen@oracle.com>
Cc: linux-kernel@vger.kernel.org, linux-scsi@vger.kernel.org,
"James E.J. Bottomley" <jejb@linux.ibm.com>,
stable@vger.kernel.org
Subject: Re: [PATCH] scsi: Let scsi_execute_cmd() mark args->sshdr as invalid
Date: Mon, 22 May 2023 17:48:50 -0500 [thread overview]
Message-ID: <b3939ccc-0da2-9156-551b-dce6ee3e8b5a@oracle.com> (raw)
In-Reply-To: <ffb8237c-bd38-e323-0179-4313d9ef0a75@oracle.com>
On 5/22/23 10:54 AM, John Garry wrote:
> On 22/05/2023 14:31, Bart Van Assche wrote:
>> On 5/22/23 02:55, John Garry wrote:
>>> On 19/05/2023 18:39, Bart Van Assche wrote:
>>>> *args->resid = scmd->resid_len;
>>>> - if (args->sense)
>>>> - memcpy(args->sense, scmd->sense_buffer, SCSI_SENSE_BUFFERSIZE);
>>>> + if (args->sense) {
>>>> + *args->sense = scmd->sense_buffer;
>>>> + scmd->sense_buffer = NULL;
>>>
>>> I think that you will agree that this is not a good pattern to follow. We cannot have SCSI core allocating the sense buffer but a driver freeing it.
>>
>> Why not? Something similar can happen anywhere in the kernel anywhere reference counting is used.
>
> Sure, but you are not using ref counting. If you could use ref counting then it would be better.
>
What about killing the sense buffer arg and doing a callback?
For the retries patchset, one issue we had was scsi_execute_cmd callers for
the most part just wanted to check different sense/asc/ascq codes. However,
there are several places that want to do something more advanced and that's
specific to their use. For them, Martin W and I had talked about a callback.
For this sense case, the callback can look at the sense buffer scsi-ml creates
for all cmds in scsi_mq_init_request, and just copy the values it wants to copy
like in ata_task_ioctl. Something like
scsi_check_passthrough()
...
if (scsi_cmnd->failures->check_failure)
scsi_cmnd->failures->check_failure(scmd, &sshdr)
ata_task_ioctl()
struct scsi_failure *failures = {
.check_failure = ata_task_check_failure,
.check_args = args,
....
bool ata_task_check_failure(struct scsi_cmnd *cmd)
{
u8 *args = scsi_cmnd->failures->check_args;
u8 *sense = cmd->sensebuf;
.....
if (scsi_sense_valid(&sshdr)) {/* sense data available */
u8 *desc = sensebuf + 8;
/* If we set cc then ATA pass-through will cause a
* check condition even if no error. Filter that. */
if (cmd_result & SAM_STAT_CHECK_CONDITION) {
if (sshdr.sense_key == RECOVERED_ERROR &&
sshdr.asc == 0 && sshdr.ascq == 0x1d)
cmd_result &= ~SAM_STAT_CHECK_CONDITION;
}
/* Send userspace ATA registers */
if (sensebuf[0] == 0x72 && /* format is "descriptor" */
desc[0] == 0x09) {/* code is "ATA Descriptor" */
args[0] = desc[13]; /* status */
args[1] = desc[3]; /* error */
args[2] = desc[5]; /* sector count (0:7) */
args[3] = desc[7]; /* lbal */
args[4] = desc[9]; /* lbam */
args[5] = desc[11]; /* lbah */
args[6] = desc[12]; /* select */
}
}
}
next prev parent reply other threads:[~2023-05-22 22:50 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-05-11 12:34 [PATCH] scsi: Let scsi_execute_cmd() mark args->sshdr as invalid Juergen Gross
2023-05-11 12:49 ` Bart Van Assche
2023-05-11 12:54 ` Juergen Gross
2023-05-11 13:10 ` Martin Wilck
2023-05-11 13:17 ` Juergen Gross
2023-05-11 13:23 ` Martin Wilck
2023-05-11 13:32 ` Juergen Gross
2023-05-11 15:59 ` Martin Wilck
2023-05-11 16:00 ` Martin Wilck
2023-05-17 2:06 ` Martin K. Petersen
2023-05-17 4:54 ` Juergen Gross
2023-05-17 15:05 ` John Garry
2023-05-18 4:53 ` Juergen Gross
2023-05-18 10:57 ` John Garry
2023-05-18 19:54 ` Bart Van Assche
2023-05-19 16:06 ` John Garry
2023-05-19 16:54 ` Bart Van Assche
2023-05-19 17:12 ` John Garry
2023-05-19 17:39 ` Bart Van Assche
2023-05-22 9:55 ` John Garry
2023-05-22 13:31 ` Bart Van Assche
2023-05-22 15:54 ` John Garry
2023-05-22 22:48 ` michael.christie [this message]
2023-05-21 1:19 ` Martin K. Petersen
2023-05-21 5:23 ` Juergen Gross
2023-05-22 22:26 ` Martin K. Petersen
2023-05-23 15:04 ` Mike Christie
2023-05-21 0:46 ` Martin K. Petersen
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=b3939ccc-0da2-9156-551b-dce6ee3e8b5a@oracle.com \
--to=michael.christie@oracle.com \
--cc=bvanassche@acm.org \
--cc=jejb@linux.ibm.com \
--cc=jgross@suse.com \
--cc=john.g.garry@oracle.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=martin.petersen@oracle.com \
--cc=stable@vger.kernel.org \
/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