linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
To: linux-nvme@lists.infradead.org
Cc: Sagi Grimberg <sagi@grimberg.me>, Jens Axboe <axboe@fb.com>,
	Keith Busch <keith.busch@intel.com>,
	Christoph Hellwig <hch@lst.de>,
	linux-kernel@vger.kernel.org, Paul Pawlowski <paul@mrarm.io>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>
Subject: [PATCH v4 4/4] nvme-pci: Support shared tags across queues for Apple 2018 controllers
Date: Wed,  7 Aug 2019 17:51:22 +1000	[thread overview]
Message-ID: <20190807075122.6247-5-benh@kernel.crashing.org> (raw)
In-Reply-To: <20190807075122.6247-1-benh@kernel.crashing.org>

Another issue with the Apple T2 based 2018 controllers seem to be
that they blow up (and shut the machine down) if there's a tag
collision between the IO queue and the Admin queue.

My suspicion is that they use our tags for their internal tracking
and don't mix them with the queue id. They also seem to not like
when tags go beyond the IO queue depth, ie 128 tags.

This adds a quirk that marks tags 0..31 of the IO queue reserved

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Acked-by: Keith Busch <keith.busch@intel.com>
---
 drivers/nvme/host/nvme.h |  5 +++++
 drivers/nvme/host/pci.c  | 31 ++++++++++++++++++++++++++++++-
 2 files changed, 35 insertions(+), 1 deletion(-)

diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index 0925f7fc13ff..3e64f7187e70 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -102,6 +102,11 @@ enum nvme_quirks {
 	 * Use non-standard 128 bytes SQEs.
 	 */
 	NVME_QUIRK_128_BYTES_SQES		= (1 << 11),
+
+	/*
+	 * Prevent tag overlap between queues
+	 */
+	NVME_QUIRK_SHARED_TAGS                  = (1 << 12),
 };
 
 /*
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index c683263cdf60..de8c170d5abc 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -2106,6 +2106,14 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
 	unsigned long size;
 
 	nr_io_queues = max_io_queues();
+
+	/*
+	 * If tags are shared with admin queue (Apple bug), then
+	 * make sure we only use one IO queue.
+	 */
+	if (dev->ctrl.quirks & NVME_QUIRK_SHARED_TAGS)
+		nr_io_queues = 1;
+
 	result = nvme_set_queue_count(&dev->ctrl, &nr_io_queues);
 	if (result < 0)
 		return result;
@@ -2276,6 +2284,14 @@ static int nvme_dev_add(struct nvme_dev *dev)
 		dev->tagset.flags = BLK_MQ_F_SHOULD_MERGE;
 		dev->tagset.driver_data = dev;
 
+		/*
+		 * Some Apple controllers requires tags to be unique
+		 * across admin and IO queue, so reserve the first 32
+		 * tags of the IO queue.
+		 */
+		if (dev->ctrl.quirks & NVME_QUIRK_SHARED_TAGS)
+			dev->tagset.reserved_tags = NVME_AQ_DEPTH;
+
 		ret = blk_mq_alloc_tag_set(&dev->tagset);
 		if (ret) {
 			dev_warn(dev->ctrl.device,
@@ -2356,6 +2372,18 @@ static int nvme_pci_enable(struct nvme_dev *dev)
                         "set queue depth=%u\n", dev->q_depth);
 	}
 
+	/*
+	 * Controllers with the shared tags quirk need the IO queue to be
+	 * big enough so that we get 32 tags for the admin queue
+	 */
+	if ((dev->ctrl.quirks & NVME_QUIRK_SHARED_TAGS) &&
+	    (dev->q_depth < (NVME_AQ_DEPTH + 2))) {
+		dev->q_depth = NVME_AQ_DEPTH + 2;
+		dev_warn(dev->ctrl.device, "IO queue depth clamped to %d\n",
+			 dev->q_depth);
+	}
+
+
 	nvme_map_cmb(dev);
 
 	pci_enable_pcie_error_reporting(pdev);
@@ -3057,7 +3085,8 @@ static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_DEVICE(PCI_VENDOR_ID_APPLE, 0x2003) },
 	{ PCI_DEVICE(PCI_VENDOR_ID_APPLE, 0x2005),
 		.driver_data = NVME_QUIRK_SINGLE_VECTOR |
-				NVME_QUIRK_128_BYTES_SQES },
+				NVME_QUIRK_128_BYTES_SQES |
+				NVME_QUIRK_SHARED_TAGS },
 	{ 0, }
 };
 MODULE_DEVICE_TABLE(pci, nvme_id_table);
-- 
2.17.1


  parent reply	other threads:[~2019-08-07  7:52 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-07  7:51 [PATCH v4 0/4] nvme-pci: Support for Apple 201+ (T2 chip) Benjamin Herrenschmidt
2019-08-07  7:51 ` [PATCH v4 1/4] nvme-pci: Pass the queue to SQ_SIZE/CQ_SIZE macros Benjamin Herrenschmidt
2019-08-07  7:51 ` [PATCH v4 2/4] nvme-pci: Add support for variable IO SQ element size Benjamin Herrenschmidt
2019-08-22  0:28   ` Christoph Hellwig
2019-08-22  0:31     ` Benjamin Herrenschmidt
2019-08-22 18:02       ` Sagi Grimberg
2019-08-23  0:41         ` Benjamin Herrenschmidt
2019-08-23  2:52           ` Sagi Grimberg
2019-08-23  3:50             ` Benjamin Herrenschmidt
2019-08-07  7:51 ` [PATCH v4 3/4] nvme-pci: Add support for Apple 2018+ models Benjamin Herrenschmidt
2019-08-22  0:28   ` Christoph Hellwig
2019-08-07  7:51 ` Benjamin Herrenschmidt [this message]
2019-08-22  0:29   ` [PATCH v4 4/4] nvme-pci: Support shared tags across queues for Apple 2018 controllers Christoph Hellwig
2019-08-08 23:52 ` [PATCH v4 0/4] nvme-pci: Support for Apple 201+ (T2 chip) Sagi Grimberg

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=20190807075122.6247-5-benh@kernel.crashing.org \
    --to=benh@kernel.crashing.org \
    --cc=axboe@fb.com \
    --cc=hch@lst.de \
    --cc=keith.busch@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    --cc=paul@mrarm.io \
    --cc=sagi@grimberg.me \
    /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).