All of lore.kernel.org
 help / color / mirror / Atom feed
From: Keith Busch <keith.busch@intel.com>
To: Bjorn Helgaas <bhelgaas@google.com>
Cc: Alexander Gordeev <agordeev@redhat.com>,
	Matthew Wilcox <willy@linux.intel.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	linux-nvme@lists.infradead.org,
	"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>
Subject: Re: [PATCH v2 5/9] nvme: Fix invalid call to irq_set_affinity_hint()
Date: Fri, 17 Jan 2014 15:01:10 -0700 (MST)	[thread overview]
Message-ID: <alpine.LRH.2.03.1401171419370.25844@AMR> (raw)
In-Reply-To: <CAErSpo4g_xm4vZWZDgyjtJu+z5ARSerQLrtJ6W_peXv4bKwcgg@mail.gmail.com>

On Fri, 17 Jan 2014, Bjorn Helgaas wrote:

> On Fri, Jan 17, 2014 at 9:02 AM, Alexander Gordeev <agordeev@redhat.com> wrote:
>> In case MSI-X and MSI initialization failed the function
>> irq_set_affinity_hint() is called with uninitialized value
>> in dev->entry[0].vector. This update fixes the issue.
>
> dev->entry[0].vector is initialized in nvme_dev_map(), and it's used
> for free_irq() above the area of your patch, so I don't think this is
> actually a bug, though it might be somewhat confusing.

It is confusing, but there's a reason. :)

We send a single command using legacy irq to discover how many msix
vectors we want. The legacy entry needs to be set some time before calling
request_irq in nvme_configure_admin_queue, but also within nvme_dev_start
(for power-management). I don't think there's a place to set it that
won't look odd when looking at nvme_setup_io_queues. I settled on
'nvme_dev_map' was because 'nvme_dev_unmap' invalidates the entries,
so this seemed to provide some amount of symmetry.

>> Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
>> ---
>>  drivers/block/nvme-core.c |   12 ++++++------
>>  1 files changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c
>> index 26d03fa..e292450 100644
>> --- a/drivers/block/nvme-core.c
>> +++ b/drivers/block/nvme-core.c
>> @@ -1790,15 +1790,15 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
>>                         vecs = 32;
>>                 for (;;) {
>>                         result = pci_enable_msi_block(pdev, vecs);
>> -                       if (result == 0) {
>> -                               for (i = 0; i < vecs; i++)
>> -                                       dev->entry[i].vector = i + pdev->irq;
>> -                               break;
>> +                       if (result > 0) {
>> +                               vecs = result;
>> +                               continue;
>>                         } else if (result < 0) {
>>                                 vecs = 1;
>> -                               break;
>>                         }
>> -                       vecs = result;
>> +                       for (i = 0; i < vecs; i++)
>> +                               dev->entry[i].vector = i + pdev->irq;
>> +                       break;
>>                 }
>>         }

WARNING: multiple messages have this Message-ID (diff)
From: keith.busch@intel.com (Keith Busch)
Subject: [PATCH v2 5/9] nvme: Fix invalid call to irq_set_affinity_hint()
Date: Fri, 17 Jan 2014 15:01:10 -0700 (MST)	[thread overview]
Message-ID: <alpine.LRH.2.03.1401171419370.25844@AMR> (raw)
In-Reply-To: <CAErSpo4g_xm4vZWZDgyjtJu+z5ARSerQLrtJ6W_peXv4bKwcgg@mail.gmail.com>

On Fri, 17 Jan 2014, Bjorn Helgaas wrote:

> On Fri, Jan 17, 2014@9:02 AM, Alexander Gordeev <agordeev@redhat.com> wrote:
>> In case MSI-X and MSI initialization failed the function
>> irq_set_affinity_hint() is called with uninitialized value
>> in dev->entry[0].vector. This update fixes the issue.
>
> dev->entry[0].vector is initialized in nvme_dev_map(), and it's used
> for free_irq() above the area of your patch, so I don't think this is
> actually a bug, though it might be somewhat confusing.

It is confusing, but there's a reason. :)

We send a single command using legacy irq to discover how many msix
vectors we want. The legacy entry needs to be set some time before calling
request_irq in nvme_configure_admin_queue, but also within nvme_dev_start
(for power-management). I don't think there's a place to set it that
won't look odd when looking at nvme_setup_io_queues. I settled on
'nvme_dev_map' was because 'nvme_dev_unmap' invalidates the entries,
so this seemed to provide some amount of symmetry.

>> Signed-off-by: Alexander Gordeev <agordeev at redhat.com>
>> ---
>>  drivers/block/nvme-core.c |   12 ++++++------
>>  1 files changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c
>> index 26d03fa..e292450 100644
>> --- a/drivers/block/nvme-core.c
>> +++ b/drivers/block/nvme-core.c
>> @@ -1790,15 +1790,15 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
>>                         vecs = 32;
>>                 for (;;) {
>>                         result = pci_enable_msi_block(pdev, vecs);
>> -                       if (result == 0) {
>> -                               for (i = 0; i < vecs; i++)
>> -                                       dev->entry[i].vector = i + pdev->irq;
>> -                               break;
>> +                       if (result > 0) {
>> +                               vecs = result;
>> +                               continue;
>>                         } else if (result < 0) {
>>                                 vecs = 1;
>> -                               break;
>>                         }
>> -                       vecs = result;
>> +                       for (i = 0; i < vecs; i++)
>> +                               dev->entry[i].vector = i + pdev->irq;
>> +                       break;
>>                 }
>>         }

  reply	other threads:[~2014-01-17 22:01 UTC|newest]

Thread overview: 114+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-01-17 16:02 [PATCH v2 0/9] Phase out pci_enable_msi_block() Alexander Gordeev
2014-01-17 16:02 ` Alexander Gordeev
2014-01-17 16:02 ` Alexander Gordeev
2014-01-17  9:26 ` [PATCH 10/9] PCI/MSI: Undeprecate pci_enable_msi() Alexander Gordeev
2014-01-29 22:10   ` Bjorn Helgaas
2014-01-17  9:27 ` [PATCH 11/9] PCI/MSI: Phase out pci_enable_msi_block() Alexander Gordeev
2014-01-29 22:22   ` Bjorn Helgaas
2014-01-17 16:02 ` [PATCH v2 1/9] ahci: Fix broken fallback to single MSI mode Alexander Gordeev
2014-01-17 16:02 ` [PATCH v2 2/9] ahci: Use pci_enable_msi_range() Alexander Gordeev
2014-01-17 16:02 ` [PATCH v2 3/9] ipr: Get rid of superfluous call to pci_disable_msi/msix() Alexander Gordeev
2014-01-17 16:02 ` [PATCH v2 4/9] ipr: Use pci_enable_msi_range() and pci_enable_msix_range() Alexander Gordeev
2014-01-28  9:54   ` Alexander Gordeev
2014-01-28 15:28     ` Brian King
2014-01-29 13:26       ` Alexander Gordeev
2014-01-30 14:06         ` Alexander Gordeev
2014-01-30 14:09           ` [PATCH v2 1/2] ipr: Get rid of superfluous call to pci_disable_msi/msix() Alexander Gordeev
2014-02-03 15:20             ` Brian King
2014-01-30 14:10           ` [PATCH v2 2/2] ipr: Use pci_enable_msi_range() and pci_enable_msix_range() Alexander Gordeev
2014-02-03 15:22             ` Brian King
2014-02-24  8:12           ` [PATCH v2 4/9] " Alexander Gordeev
2014-02-26 15:02             ` Brian King
2014-01-17 16:02 ` [PATCH v2 5/9] nvme: Fix invalid call to irq_set_affinity_hint() Alexander Gordeev
2014-01-17 16:02   ` Alexander Gordeev
2014-01-17 19:40   ` Bjorn Helgaas
2014-01-17 19:40     ` Bjorn Helgaas
2014-01-17 22:01     ` Keith Busch [this message]
2014-01-17 22:01       ` Keith Busch
2014-01-20  8:38       ` Alexander Gordeev
2014-01-20  8:38         ` Alexander Gordeev
2014-01-20  8:40         ` [PATCH v3 6/9] nvme: Use pci_enable_msi_range() and pci_enable_msix_range() Alexander Gordeev
2014-01-20  8:40           ` Alexander Gordeev
2014-02-05 13:07           ` Alexander Gordeev
2014-02-05 13:07             ` Alexander Gordeev
2014-02-18 17:54           ` Keith Busch
2014-02-18 17:54             ` Keith Busch
2014-01-20  8:42         ` [PATCH] nvme: Cleanup nvme_dev_start() and fix IRQ leak Alexander Gordeev
2014-01-20  8:42           ` Alexander Gordeev
2014-01-20 22:48           ` Keith Busch
2014-01-20 22:48             ` Keith Busch
2014-01-21 10:03             ` Alexander Gordeev
2014-01-21 10:03               ` Alexander Gordeev
2014-01-21 10:06               ` [PATCH 1/2] Revert "NVMe: Disable admin queue on init failure" Alexander Gordeev
2014-01-21 10:06                 ` Alexander Gordeev
2014-01-21 10:07               ` [PATCH 2/2] nvme: Cleanup nvme_dev_start() and fix IRQ leak Alexander Gordeev
2014-01-21 10:07                 ` Alexander Gordeev
2014-01-21 19:06                 ` Keith Busch
2014-01-21 19:06                   ` Keith Busch
2014-01-20  8:43         ` [PATCH] nvme: Cleanup nvme_dev_start() Alexander Gordeev
2014-01-20  8:43           ` Alexander Gordeev
2014-01-20 16:41           ` Keith Busch
2014-01-20 16:41             ` Keith Busch
2014-01-17 16:02 ` [PATCH v2 6/9] nvme: Use pci_enable_msi_range() and pci_enable_msix_range() Alexander Gordeev
2014-01-17 16:02   ` Alexander Gordeev
2014-01-17 16:02 ` [PATCH v2 7/9] vfio: " Alexander Gordeev
2014-01-29 21:36   ` Bjorn Helgaas
2014-01-17 16:02 ` [PATCH v2 8/9] ath10k: Use pci_enable_msi_range() Alexander Gordeev
2014-01-17 16:02   ` Alexander Gordeev
2014-01-30 13:48   ` [PATCH 1/3] ath10k: Get rid of superfluous call to pci_disable_msi() Alexander Gordeev
2014-01-30 13:48     ` Alexander Gordeev
2014-02-04 18:32     ` Kalle Valo
2014-02-04 18:32       ` Kalle Valo
2014-02-04 19:09       ` Alexander Gordeev
2014-02-04 19:09         ` Alexander Gordeev
2014-02-05  8:21         ` Kalle Valo
2014-02-05  8:21           ` Kalle Valo
2014-02-05  8:50           ` Alexander Gordeev
2014-02-05  8:50             ` Alexander Gordeev
2014-02-05  8:54             ` Kalle Valo
2014-02-05  8:54               ` Kalle Valo
2014-02-12  0:31               ` Bjorn Helgaas
2014-02-12  0:31                 ` Bjorn Helgaas
2014-02-12 13:38                 ` Alexander Gordeev
2014-02-12 13:38                   ` Alexander Gordeev
2014-02-12 19:28                   ` Bjorn Helgaas
2014-02-12 19:28                     ` Bjorn Helgaas
2014-02-12 21:30                     ` Kalle Valo
2014-02-12 21:30                       ` Kalle Valo
2014-02-12 21:40                       ` Bjorn Helgaas
2014-02-12 21:40                         ` Bjorn Helgaas
2014-02-13 10:29                         ` Kalle Valo
2014-02-13 10:29                           ` Kalle Valo
2014-02-13 13:18                       ` Alexander Gordeev
2014-02-13 16:04     ` Kalle Valo
2014-02-13 16:04       ` Kalle Valo
2014-01-30 13:48   ` [PATCH 2/3] ath10k: Disable MSI in case IRQ configuration is unknown Alexander Gordeev
2014-01-30 13:48     ` Alexander Gordeev
2014-01-30 13:49   ` [PATCH v3 3/3] ath10k: Use pci_enable_msi_range() Alexander Gordeev
2014-01-30 13:49     ` Alexander Gordeev
2014-02-03 11:02   ` [PATCH v2 8/9] " Alexander Gordeev
2014-02-03 11:02     ` Alexander Gordeev
2014-01-17 16:02 ` [PATCH v2 9/9] wil6210: " Alexander Gordeev
2014-01-29 21:37   ` Bjorn Helgaas
2014-02-07 14:46   ` [PATCH] wil6210: Fix switch operator "missing break?" warning Alexander Gordeev
2014-02-10 10:54     ` Vladimir Kondratiev
2014-02-10 12:16       ` Alexander Gordeev
2014-02-11  1:32         ` Fengguang Wu
2014-01-17 21:00 ` [PATCH v2 0/9] Phase out pci_enable_msi_block() Bjorn Helgaas
2014-01-17 21:00   ` Bjorn Helgaas
2014-01-17 21:00   ` Bjorn Helgaas
2014-01-18  7:15   ` Alexander Gordeev
2014-01-18  7:15     ` Alexander Gordeev
2014-01-18  7:15     ` Alexander Gordeev
2014-01-18 14:38     ` Bjorn Helgaas
2014-01-18 14:38       ` Bjorn Helgaas
2014-01-18 14:38       ` Bjorn Helgaas
2014-01-18 14:59       ` Tejun Heo
2014-01-18 14:59         ` Tejun Heo
2014-01-18 14:59         ` Tejun Heo
2014-01-29 21:48         ` Bjorn Helgaas
2014-01-29 21:48           ` Bjorn Helgaas
2014-01-29 21:48           ` Bjorn Helgaas
2014-01-29 13:59   ` Alexander Gordeev
2014-01-29 13:59     ` Alexander Gordeev
2014-01-29 13:59     ` Alexander Gordeev

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=alpine.LRH.2.03.1401171419370.25844@AMR \
    --to=keith.busch@intel.com \
    --cc=agordeev@redhat.com \
    --cc=bhelgaas@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=willy@linux.intel.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.