Linux MIPS Architecture development
 help / color / mirror / Atom feed
From: James Hogan <james.hogan@imgtec.com>
To: Ralf Baechle <ralf@linux-mips.org>
Cc: James Hogan <james.hogan@imgtec.com>,
	Paul Burton <paul.burton@imgtec.com>,
	Leonid Yegoshin <leonid.yegoshin@imgtec.com>,
	<linux-mips@linux-mips.org>
Subject: [PATCH 13/13] MIPS: c-r4k: Use SMP calls for CM indexed cache ops
Date: Wed, 13 Jul 2016 14:12:56 +0100	[thread overview]
Message-ID: <1468415576-12600-14-git-send-email-james.hogan@imgtec.com> (raw)
In-Reply-To: <1468415576-12600-1-git-send-email-james.hogan@imgtec.com>

The MIPS Coherence Manager (CM) can propagate address-based ("hit")
cache operations to other cores in the coherent system, alleviating
software of the need to use SMP calls, however indexed cache operations
are not propagated by hardware since doing so makes no sense for
separate caches.

Update r4k_op_needs_ipi() to report that only hit cache operations are
globalized by the CM, requiring indexed cache operations to be
globalized by software via an SMP call.

r4k_on_each_cpu() previously had a special case for CONFIG_MIPS_MT_SMP,
intended to avoid the SMP calls when the only other CPUs in the system
were other VPEs in the same core, and hence sharing the same caches.
This was changed by commit cccf34e9411c ("MIPS: c-r4k: Fix cache
flushing for MT cores") to apparently handle multi-core multi-VPE
systems, but it focussed mainly on hit cache ops, so the SMP calls were
still disabled entirely for CM systems.

This doesn't normally cause problems, but tests can be written to hit
these corner cases by using multiple threads, or changing task
affinities to force the process to migrate cores. For example the
failure of mprotect RW->RX to globally sync icaches (via
flush_cache_range) can be detected by modifying and mprotecting a code
page on one core, and migrating to a different core to execute from it.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: Leonid Yegoshin <leonid.yegoshin@imgtec.com>
Cc: linux-mips@linux-mips.org
---
 arch/mips/mm/c-r4k.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/mips/mm/c-r4k.c b/arch/mips/mm/c-r4k.c
index 8016babe5c84..3227a0a267f5 100644
--- a/arch/mips/mm/c-r4k.c
+++ b/arch/mips/mm/c-r4k.c
@@ -68,7 +68,7 @@
 static inline bool r4k_op_needs_ipi(unsigned int type)
 {
 	/* The MIPS Coherence Manager (CM) globalizes address-based cache ops */
-	if (mips_cm_present())
+	if (type == R4K_HIT && mips_cm_present())
 		return false;
 
 	/*
-- 
2.4.10

WARNING: multiple messages have this Message-ID (diff)
From: James Hogan <james.hogan@imgtec.com>
To: Ralf Baechle <ralf@linux-mips.org>
Cc: James Hogan <james.hogan@imgtec.com>,
	Paul Burton <paul.burton@imgtec.com>,
	Leonid Yegoshin <leonid.yegoshin@imgtec.com>,
	linux-mips@linux-mips.org
Subject: [PATCH 13/13] MIPS: c-r4k: Use SMP calls for CM indexed cache ops
Date: Wed, 13 Jul 2016 14:12:56 +0100	[thread overview]
Message-ID: <1468415576-12600-14-git-send-email-james.hogan@imgtec.com> (raw)
Message-ID: <20160713131256.GSmcJHfWqz2IE3MsbaMhRZwBgXtcm7CzM62jtKHuGKc@z> (raw)
In-Reply-To: <1468415576-12600-1-git-send-email-james.hogan@imgtec.com>

The MIPS Coherence Manager (CM) can propagate address-based ("hit")
cache operations to other cores in the coherent system, alleviating
software of the need to use SMP calls, however indexed cache operations
are not propagated by hardware since doing so makes no sense for
separate caches.

Update r4k_op_needs_ipi() to report that only hit cache operations are
globalized by the CM, requiring indexed cache operations to be
globalized by software via an SMP call.

r4k_on_each_cpu() previously had a special case for CONFIG_MIPS_MT_SMP,
intended to avoid the SMP calls when the only other CPUs in the system
were other VPEs in the same core, and hence sharing the same caches.
This was changed by commit cccf34e9411c ("MIPS: c-r4k: Fix cache
flushing for MT cores") to apparently handle multi-core multi-VPE
systems, but it focussed mainly on hit cache ops, so the SMP calls were
still disabled entirely for CM systems.

This doesn't normally cause problems, but tests can be written to hit
these corner cases by using multiple threads, or changing task
affinities to force the process to migrate cores. For example the
failure of mprotect RW->RX to globally sync icaches (via
flush_cache_range) can be detected by modifying and mprotecting a code
page on one core, and migrating to a different core to execute from it.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: Leonid Yegoshin <leonid.yegoshin@imgtec.com>
Cc: linux-mips@linux-mips.org
---
 arch/mips/mm/c-r4k.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/mips/mm/c-r4k.c b/arch/mips/mm/c-r4k.c
index 8016babe5c84..3227a0a267f5 100644
--- a/arch/mips/mm/c-r4k.c
+++ b/arch/mips/mm/c-r4k.c
@@ -68,7 +68,7 @@
 static inline bool r4k_op_needs_ipi(unsigned int type)
 {
 	/* The MIPS Coherence Manager (CM) globalizes address-based cache ops */
-	if (mips_cm_present())
+	if (type == R4K_HIT && mips_cm_present())
 		return false;
 
 	/*
-- 
2.4.10

  parent reply	other threads:[~2016-07-13 13:18 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-07-13 13:12 [PATCH 00/13] MIPS: c-r4k: SMP cache management fixes James Hogan
2016-07-13 13:12 ` James Hogan
2016-07-13 13:12 ` [PATCH 01/13] MIPS: SMP: Clear ASID without confusing has_valid_asid() James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 02/13] MIPS: SMP: Update cpu_foreign_map on CPU disable James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 03/13] MIPS: SMP: Drop stop_this_cpu() cpu_foreign_map hack James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 04/13] MIPS: c-r4k: Fix protected_writeback_scache_line for EVA James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 05/13] MIPS: c-r4k: Fix sigtramp SMP call to use kmap James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 06/13] MIPS: c-r4k: Avoid dcache flush for sigtramps James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 07/13] MIPS: c-r4k: Add r4k_on_each_cpu cache op type arg James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 08/13] MIPS: c-r4k: Fix valid ASID optimisation James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 09/13] MIPS: c-r4k: Exclude sibling CPUs in SMP calls James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 10/13] MIPS: c-r4k: Split r4k_flush_kernel_vmap_range() James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 11/13] MIPS: c-r4k: Local flush_icache_range cache op override James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` [PATCH 12/13] MIPS: c-r4k: Avoid small flush_icache_range SMP calls James Hogan
2016-07-13 13:12   ` James Hogan
2016-07-13 13:12 ` James Hogan [this message]
2016-07-13 13:12   ` [PATCH 13/13] MIPS: c-r4k: Use SMP calls for CM indexed cache ops James Hogan

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=1468415576-12600-14-git-send-email-james.hogan@imgtec.com \
    --to=james.hogan@imgtec.com \
    --cc=leonid.yegoshin@imgtec.com \
    --cc=linux-mips@linux-mips.org \
    --cc=paul.burton@imgtec.com \
    --cc=ralf@linux-mips.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