Linux IOMMU Development
 help / color / mirror / Atom feed
From: Scott Wood <swood-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
To: Joerg Roedel <joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org>
Cc: "Luis Claudio R. Goncalves"
	<lgoncalv-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
	Clark Williams <williams-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
	iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org,
	Scott Wood <swood-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Subject: [PATCH v2] iommu/amd: Avoid locking get_irq_table() from atomic context
Date: Wed, 14 Feb 2018 17:36:28 -0600	[thread overview]
Message-ID: <20180214233628.29560-1-swood@redhat.com> (raw)

get_irq_table() previously acquired amd_iommu_devtable_lock which is not
a raw lock, and thus cannot be acquired from atomic context on
PREEMPT_RT.  Many calls to modify_irte*() come from atomic context due to
the IRQ desc->lock, as does amd_iommu_update_ga() due to the preemption
disabling in vcpu_load/put().

The only difference between calling get_irq_table() and reading from
irq_lookup_table[] directly, other than the lock acquisition and
amd_iommu_rlookup_table[] check, is if the table entry is unpopulated,
which should never happen when looking up a devid that came from an
irq_2_irte struct, as get_irq_table() would have already been called on
that devid during irq_remapping_alloc().

The lock acquisition is not needed in these cases because entries in
irq_lookup_table[] never change once non-NULL -- nor would the
amd_iommu_devtable_lock usage in get_irq_table() provide meaningful
protection if they did, since it's released before using the looked up
table in the get_irq_table() caller.

Rename the old get_irq_table() to alloc_irq_table(), and create a new
lockless get_irq_table() to be used in non-allocating contexts that WARNs
if it doesn't find what it's looking for.

Signed-off-by: Scott Wood <swood-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
v2: Added new get_irq_table() with WARNs rather than accessing
irq_lookup_table[] directly.
---
 drivers/iommu/amd_iommu.c | 29 ++++++++++++++++++++++-------
 1 file changed, 22 insertions(+), 7 deletions(-)

diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
index e4026133aa1d..5d41e0733cb3 100644
--- a/drivers/iommu/amd_iommu.c
+++ b/drivers/iommu/amd_iommu.c
@@ -3594,7 +3594,22 @@ static void set_dte_irq_entry(u16 devid, struct irq_remap_table *table)
 	amd_iommu_dev_table[devid].data[2] = dte;
 }
 
-static struct irq_remap_table *get_irq_table(u16 devid, bool ioapic)
+static struct irq_remap_table *get_irq_table(u16 devid)
+{
+	struct irq_remap_table *table;
+
+	if (WARN_ONCE(!amd_iommu_rlookup_table[devid],
+		      "%s: no iommu for devid %x\n", __func__, devid))
+		return NULL;
+
+	table = irq_lookup_table[devid];
+	if (WARN_ONCE(!table, "%s: no table for devid %x\n", __func__, devid))
+		return NULL;
+
+	return table;
+}
+
+static struct irq_remap_table *alloc_irq_table(u16 devid, bool ioapic)
 {
 	struct irq_remap_table *table = NULL;
 	struct amd_iommu *iommu;
@@ -3681,7 +3696,7 @@ static int alloc_irq_index(u16 devid, int count, bool align)
 	if (!iommu)
 		return -ENODEV;
 
-	table = get_irq_table(devid, false);
+	table = alloc_irq_table(devid, false);
 	if (!table)
 		return -ENODEV;
 
@@ -3732,7 +3747,7 @@ static int modify_irte_ga(u16 devid, int index, struct irte_ga *irte,
 	if (iommu == NULL)
 		return -EINVAL;
 
-	table = get_irq_table(devid, false);
+	table = get_irq_table(devid);
 	if (!table)
 		return -ENOMEM;
 
@@ -3765,7 +3780,7 @@ static int modify_irte(u16 devid, int index, union irte *irte)
 	if (iommu == NULL)
 		return -EINVAL;
 
-	table = get_irq_table(devid, false);
+	table = get_irq_table(devid);
 	if (!table)
 		return -ENOMEM;
 
@@ -3789,7 +3804,7 @@ static void free_irte(u16 devid, int index)
 	if (iommu == NULL)
 		return;
 
-	table = get_irq_table(devid, false);
+	table = get_irq_table(devid);
 	if (!table)
 		return;
 
@@ -4107,7 +4122,7 @@ static int irq_remapping_alloc(struct irq_domain *domain, unsigned int virq,
 		return ret;
 
 	if (info->type == X86_IRQ_ALLOC_TYPE_IOAPIC) {
-		if (get_irq_table(devid, true))
+		if (alloc_irq_table(devid, true))
 			index = info->ioapic_pin;
 		else
 			ret = -ENOMEM;
@@ -4390,7 +4405,7 @@ int amd_iommu_update_ga(int cpu, bool is_run, void *data)
 	if (!iommu)
 		return -ENODEV;
 
-	irt = get_irq_table(devid, false);
+	irt = get_irq_table(devid);
 	if (!irt)
 		return -ENODEV;
 
-- 
2.14.3

             reply	other threads:[~2018-02-14 23:36 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-02-14 23:36 Scott Wood [this message]
     [not found] ` <20180214233628.29560-1-swood-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2018-02-15 11:04   ` [PATCH v2] iommu/amd: Avoid locking get_irq_table() from atomic context Joerg Roedel

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=20180214233628.29560-1-swood@redhat.com \
    --to=swood-h+wxahxf7alqt0dzr+alfa@public.gmane.org \
    --cc=iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org \
    --cc=joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org \
    --cc=lgoncalv-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
    --cc=williams-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.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