From: Damien Le Moal <Damien.LeMoal@wdc.com>
To: "Reimar Döffinger" <Reimar.Doeffinger@gmx.de>,
"linux-ide@vger.kernel.org" <linux-ide@vger.kernel.org>,
"Jens Axboe" <axboe@kernel.dk>,
"hch@infradead.org" <hch@infradead.org>
Subject: Re: [PATCH] libata: Disable ATA_CMD_READ_LOG_DMA_EXT in problematic cases.
Date: Mon, 16 Aug 2021 00:10:28 +0000 [thread overview]
Message-ID: <DM6PR04MB708117F3FD8537CB6875E2C4E7FD9@DM6PR04MB7081.namprd04.prod.outlook.com> (raw)
In-Reply-To: 20210815162725.15124-1-Reimar.Doeffinger@gmx.de
On 2021/08/16 1:27, Reimar Döffinger wrote:
> The ATA_CMD_READ_LOG_DMA_EXT can cause controller/device to
> become unresponsive until the next power cycle.
> This seems to particularly affect IDE to SATA adapters,
> possibly in combination with certain SATA SSDs, though
> there might be more/different cases.
> Comment 5 of https://bugzilla.kernel.org/show_bug.cgi?id=195895
> is an example.
> Disable it by default on Crucial MX500 devices and all
> PATA controllers.
> Also add an option to set the workaround state, which might
> help with gathering more data on the exact devices/controllers
> affected, and speed up narrowing down the cause for users that
> are affect but not covered by the added quirks.
> Existing workarounds like forcing PIO mode do not work
> (in addition to the performance issues) because READ_LOG_DMA
> is issued even if PIO mode is forced.
>
> Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
> ---
> Documentation/admin-guide/kernel-parameters.txt | 2 ++
> drivers/ata/libata-core.c | 15 +++++++++++++++
> 2 files changed, 17 insertions(+)
>
> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> index bdb22006f713..191502e8fa74 100644
> --- a/Documentation/admin-guide/kernel-parameters.txt
> +++ b/Documentation/admin-guide/kernel-parameters.txt
> @@ -2551,6 +2551,8 @@
>
> * [no]ncqtrim: Turn off queued DSM TRIM.
>
> + * [no]dmalog: Turn off use of ATA_CMD_READ_LOG_DMA_EXT (0x47) command
> +
> * nohrst, nosrst, norst: suppress hard, soft
> and both resets.
>
> diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
> index 61c762961ca8..219fa92ffc06 100644
> --- a/drivers/ata/libata-core.c
> +++ b/drivers/ata/libata-core.c
> @@ -2004,7 +2004,11 @@ unsigned int ata_read_log_page(struct ata_device *dev, u8 log,
>
> retry:
> ata_tf_init(dev, &tf);
> + /* Do not even attempt DMA with PATA-SATA adapters, they seem likely to
> + * hang, see https://bugzilla.kernel.org/show_bug.cgi?id=195895
> + */
This is not the standard kernel comment style. Multi-lines comment should start
with a "/*" line:
+ /*
+ * Do not even attempt DMA with PATA-SATA adapters, they seem likely to
+ * hang, see https://bugzilla.kernel.org/show_bug.cgi?id=195895
+ */
Also, I would remove the bugzilla reference since it is in the commit message.
> if (dev->dma_mode && ata_id_has_read_log_dma_ext(dev->id) &&
> + (ap_flags & ATA_FLAG_SATA) &&
Why is this necessary since you have the ATA_HORKAGE_NO_DMA_LOG flag ?
If this is really necessary, ATA_HORKAGE_NO_DMA_LOG should be set for all
devices that have ATA_FLAG_SATA when the device is initialized. With that, this
additional condition can go away.
> !(dev->horkage & ATA_HORKAGE_NO_DMA_LOG)) {
> tf.command = ATA_CMD_READ_LOG_DMA_EXT;
> tf.protocol = ATA_PROT_DMA;
> @@ -4000,6 +4004,15 @@ static const struct ata_blacklist_entry ata_device_blacklist [] = {
> { "WDC WD3000JD-*", NULL, ATA_HORKAGE_WD_BROKEN_LPM },
> { "WDC WD3200JD-*", NULL, ATA_HORKAGE_WD_BROKEN_LPM },
>
> + /* Devices with observed READ LOG DMA issues - unclear if only
> + * specific firmware versions or only in combination with specific
> + * controllers.
> + * Specifically broken combinations reported
> + * CT1000MX500SSD4, M3CR020 with B350M-Mortar
> + * CT500MX500SSD4, M3CR023 with PATA-SATA adapter
> + * https://bugzilla.kernel.org/show_bug.cgi?id=195895
> + */
Same comment about the comment format.
I would also remove the bugzilla reference as it is in the commit message.
> + { "Crucial_CT*MX500*", NULL, ATA_HORKAGE_NO_DMA_LOG },
> /* End Marker */
> { }
> };
> @@ -6104,6 +6117,8 @@ static int __init ata_parse_force_one(char **cur,
> { "ncq", .horkage_off = ATA_HORKAGE_NONCQ },
> { "noncqtrim", .horkage_on = ATA_HORKAGE_NO_NCQ_TRIM },
> { "ncqtrim", .horkage_off = ATA_HORKAGE_NO_NCQ_TRIM },
> + { "nodmalog", .horkage_on = ATA_HORKAGE_NO_DMA_LOG },
> + { "dmalog", .horkage_off = ATA_HORKAGE_NO_DMA_LOG },
> { "dump_id", .horkage_on = ATA_HORKAGE_DUMP_ID },
> { "pio0", .xfer_mask = 1 << (ATA_SHIFT_PIO + 0) },
> { "pio1", .xfer_mask = 1 << (ATA_SHIFT_PIO + 1) },
>
--
Damien Le Moal
Western Digital Research
next prev parent reply other threads:[~2021-08-16 0:10 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-14 13:47 [PATCH] libata: add kernel parameter to force ATA_HORKAGE_NO_DMA_LOG Reimar Döffinger
2021-08-15 8:18 ` Christoph Hellwig
2021-08-15 16:27 ` [PATCH] libata: Disable ATA_CMD_READ_LOG_DMA_EXT in problematic cases Reimar Döffinger
2021-08-16 0:10 ` Damien Le Moal [this message]
2021-08-16 17:15 ` Reimar Döffinger
2021-08-17 19:06 ` Reimar Döffinger
2021-08-18 22:49 ` Damien Le Moal
2021-08-16 17:26 ` Reimar Döffinger
2021-08-17 7:45 ` Damien Le Moal
2021-08-17 18:01 ` Reimar Döffinger
2021-08-17 19:03 ` [PATCH] libata: fix setting and checking of DMA state Reimar Döffinger
2021-08-18 23:30 ` Damien Le Moal
2021-08-19 8:13 ` [PATCH] libata: fix " Reimar Döffinger
2021-08-19 12:56 ` Reimar Döffinger
2021-09-27 8:56 ` Paul Menzel
2021-09-27 9:10 ` Reimar Döffinger
2021-09-27 9:15 ` Damien Le Moal
2021-10-03 13:28 ` [PATCH v4] Fixes to DMA state check Reimar Döffinger
2021-10-03 13:28 ` [PATCH 1/6] libata: fix checking of DMA state Reimar Döffinger
2021-10-03 13:28 ` [PATCH 2/6] libata-scsi: " Reimar Döffinger
2021-10-03 13:28 ` [PATCH 3/6] pata_ali: " Reimar Döffinger
2021-10-03 13:28 ` [PATCH 4/6] pata_amd: " Reimar Döffinger
2021-10-03 13:28 ` [PATCH 5/6] pata_optidma: " Reimar Döffinger
2021-10-03 13:28 ` [PATCH 6/6] pata_radisys: " Reimar Döffinger
2021-10-12 2:20 ` [PATCH v4] Fixes to DMA state check Damien Le Moal
2021-10-12 6:30 ` Reimar Döffinger
[not found] ` <605EE991-5DA2-4A8D-9691-967D68D3AB51@gmx.de>
2021-10-12 6:32 ` Damien Le Moal
2021-10-12 6:33 ` Damien Le Moal
2021-08-15 16:34 ` [PATCH] libata: add kernel parameter to force ATA_HORKAGE_NO_DMA_LOG Reimar Döffinger
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=DM6PR04MB708117F3FD8537CB6875E2C4E7FD9@DM6PR04MB7081.namprd04.prod.outlook.com \
--to=damien.lemoal@wdc.com \
--cc=Reimar.Doeffinger@gmx.de \
--cc=axboe@kernel.dk \
--cc=hch@infradead.org \
--cc=linux-ide@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;
as well as URLs for NNTP newsgroup(s).