Discussion:
[PATCH 3.12 002/182] ALSA: hda - automute via amp instead of pinctl on some AIO models
Jiri Slaby
2014-05-13 09:25:54 UTC
Permalink
From: Hui Wang <***@canonical.com>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit 493a52a9b6645f61954580c7d4bd52fa62110934 upstream.

On some AIO (All In One) models with the codec alc668
(Vendor ID: 0x10ec0668) on it, when we plug a headphone into the jack,
the system will switch the output to headphone and set the speaker to
automute as well as change the speaker Pin-ctls from 0x40 to 0x00,
this will bring loud noise to the headphone.

I tried to disable the corresponding EAPD, but it did not help to
eliminate the noise.

According to Takashi's suggestion, we use amp operation to replace the
pinctl modification for the automute, this really eliminate the noise.

BugLink: https://bugs.launchpad.net/bugs/1268468
Cc: David Henningsson <***@canonical.com>
Cc: ***@vger.kernel.org
Signed-off-by: Hui Wang <***@canonical.com>
Signed-off-by: Takashi Iwai <***@suse.de>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
sound/pci/hda/patch_realtek.c | 22 +++++++++++++++++++---
1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
index 6a32c857f704..855074950e6c 100644
--- a/sound/pci/hda/patch_realtek.c
+++ b/sound/pci/hda/patch_realtek.c
@@ -3480,6 +3480,15 @@ static void alc_fixup_no_shutup(struct hda_codec *codec,
}
}

+static void alc_fixup_auto_mute_via_amp(struct hda_codec *codec,
+ const struct hda_fixup *fix, int action)
+{
+ if (action == HDA_FIXUP_ACT_PRE_PROBE) {
+ struct alc_spec *spec = codec->spec;
+ spec->gen.auto_mute_via_amp = 1;
+ }
+}
+
static void alc_fixup_headset_mode_alc668(struct hda_codec *codec,
const struct hda_fixup *fix, int action)
{
@@ -4646,6 +4655,7 @@ enum {
ALC662_FIXUP_BASS_CHMAP,
ALC662_FIXUP_BASS_1A,
ALC662_FIXUP_BASS_1A_CHMAP,
+ ALC668_FIXUP_AUTO_MUTE,
};

static const struct hda_fixup alc662_fixups[] = {
@@ -4806,6 +4816,12 @@ static const struct hda_fixup alc662_fixups[] = {
.type = HDA_FIXUP_FUNC,
.v.func = alc_fixup_inv_dmic_0x12,
},
+ [ALC668_FIXUP_AUTO_MUTE] = {
+ .type = HDA_FIXUP_FUNC,
+ .v.func = alc_fixup_auto_mute_via_amp,
+ .chained = true,
+ .chain_id = ALC668_FIXUP_DELL_MIC_NO_PRESENCE
+ },
[ALC668_FIXUP_DELL_MIC_NO_PRESENCE] = {
.type = HDA_FIXUP_PINS,
.v.pins = (const struct hda_pintbl[]) {
@@ -4851,11 +4867,11 @@ static const struct snd_pci_quirk alc662_fixup_tbl[] = {
SND_PCI_QUIRK(0x1025, 0x038b, "Acer Aspire 8943G", ALC662_FIXUP_ASPIRE),
SND_PCI_QUIRK(0x1028, 0x05d8, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
SND_PCI_QUIRK(0x1028, 0x05db, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
- SND_PCI_QUIRK(0x1028, 0x0623, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
- SND_PCI_QUIRK(0x1028, 0x0624, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
+ SND_PCI_QUIRK(0x1028, 0x0623, "Dell", ALC668_FIXUP_AUTO_MUTE),
+ SND_PCI_QUIRK(0x1028, 0x0624, "Dell", ALC668_FIXUP_AUTO_MUTE),
SND_PCI_QUIRK(0x1028, 0x0625, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
SND_PCI_QUIRK(0x1028, 0x0626, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
- SND_PCI_QUIRK(0x1028, 0x0628, "Dell", ALC668_FIXUP_DELL_MIC_NO_PRESENCE),
+ SND_PCI_QUIRK(0x1028, 0x0628, "Dell", ALC668_FIXUP_AUTO_MUTE),
SND_PCI_QUIRK(0x103c, 0x1632, "HP RP5800", ALC662_FIXUP_HP_RP5800),
SND_PCI_QUIRK(0x1043, 0x11cd, "Asus N550", ALC662_FIXUP_BASS_1A_CHMAP),
SND_PCI_QUIRK(0x1043, 0x1477, "ASUS N56VZ", ALC662_FIXUP_BASS_CHMAP),
--
1.9.3
Jiri Slaby
2014-05-13 09:26:11 UTC
Permalink
From: Xiangyu Lu <***@huawei.com>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit 80bb3ef109ff40a7593d9481c17de9bbc4d7c0e2 upstream.

In big-endian systems, "%1" get the most significant part of the value, cause the instruction to get the wrong result.

When viewing ftrace record in big-endian ARM systems, we found that
the timestamp errors:

swapper-0 [001] 1325.970000: 0:120:R ==> [001] 16:120:R events/1
events/1-16 [001] 1325.970000: 16:120:S ==> [001] 0:120:R swapper
swapper-0 [000] 1325.1000000: 0:120:R + [000] 15:120:R events/0
swapper-0 [000] 1325.1000000: 0:120:R ==> [000] 15:120:R events/0
swapper-0 [000] 1326.030000: 0:120:R + [000] 1150:120:R sshd
swapper-0 [000] 1326.030000: 0:120:R ==> [000] 1150:120:R sshd

When viewed ftrace records, it will call the do_div(n, base) function, which achieved arch/arm/include/asm/div64.h in. When n = 10000000, base = 1000000, in do_div(n, base) will execute "umull %Q0, %R0, %1, %Q2".

Reviewed-by: Dave Martin <***@arm.com>
Reviewed-by: Nicolas Pitre <***@linaro.org>
Signed-off-by: Alex Wu <***@huawei.com>
Signed-off-by: Xiangyu Lu <***@huawei.com>
Signed-off-by: Russell King <rmk+***@arm.linux.org.uk>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
arch/arm/include/asm/div64.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm/include/asm/div64.h b/arch/arm/include/asm/div64.h
index 191ada6e4d2d..662c7bd06108 100644
--- a/arch/arm/include/asm/div64.h
+++ b/arch/arm/include/asm/div64.h
@@ -156,7 +156,7 @@
/* Select the best insn combination to perform the */ \
/* actual __m * __n / (__p << 64) operation. */ \
if (!__c) { \
- asm ( "umull %Q0, %R0, %1, %Q2\n\t" \
+ asm ( "umull %Q0, %R0, %Q1, %Q2\n\t" \
"mov %Q0, #0" \
: "=&r" (__res) \
: "r" (__m), "r" (__n) \
--
1.9.3
Jiri Slaby
2014-05-13 09:28:03 UTC
Permalink
From: James Hogan <***@imgtec.com>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit 15505679362270d02c449626385cb74af8905514 upstream.

Previously a reserved instruction exception while in guest code would
cause a KVM internal error if kvm_mips_handle_ri() didn't recognise the
instruction (including a RDHWR from an unrecognised hardware register).

However the guest OS should really have the opportunity to catch the
exception so that it can take the appropriate actions such as sending a
SIGILL to the guest user process or emulating the instruction itself.

Therefore in these cases emulate a guest RI exception and only return
EMULATE_FAIL if that fails, being careful to revert the PC first in case
the exception occurred in a branch delay slot in which case the PC will
already point to the branch target.

Also turn the printk messages relating to these cases into kvm_debug
messages so that they aren't usually visible.

This allows crashme to run in the guest without killing the entire VM.

Signed-off-by: James Hogan <***@imgtec.com>
Cc: Ralf Baechle <***@linux-mips.org>
Cc: Gleb Natapov <***@kernel.org>
Cc: Paolo Bonzini <***@redhat.com>
Cc: Sanjay Lal <***@kymasys.com>
Cc: linux-***@linux-mips.org
Cc: ***@vger.kernel.org
Signed-off-by: Paolo Bonzini <***@redhat.com>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
arch/mips/kvm/kvm_mips_emul.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/arch/mips/kvm/kvm_mips_emul.c b/arch/mips/kvm/kvm_mips_emul.c
index 4b6274b47f33..e75ef8219caf 100644
--- a/arch/mips/kvm/kvm_mips_emul.c
+++ b/arch/mips/kvm/kvm_mips_emul.c
@@ -1571,17 +1571,17 @@ kvm_mips_handle_ri(unsigned long cause, uint32_t *opc,
arch->gprs[rt] = kvm_read_c0_guest_userlocal(cop0);
#else
/* UserLocal not implemented */
- er = kvm_mips_emulate_ri_exc(cause, opc, run, vcpu);
+ er = EMULATE_FAIL;
#endif
break;

default:
- printk("RDHWR not supported\n");
+ kvm_debug("RDHWR %#x not supported @ %p\n", rd, opc);
er = EMULATE_FAIL;
break;
}
} else {
- printk("Emulate RI not supported @ %p: %#x\n", opc, inst);
+ kvm_debug("Emulate RI not supported @ %p: %#x\n", opc, inst);
er = EMULATE_FAIL;
}

@@ -1590,6 +1590,7 @@ kvm_mips_handle_ri(unsigned long cause, uint32_t *opc,
*/
if (er == EMULATE_FAIL) {
vcpu->arch.pc = curr_pc;
+ er = kvm_mips_emulate_ri_exc(cause, opc, run, vcpu);
}
return er;
}
--
1.9.3
Jiri Slaby
2014-05-13 09:27:02 UTC
Permalink
From: Mike Marciniszyn <***@intel.com>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit f8b6c47a44c063062317646683a73371c24c69ee upstream.

The debugfs init code was incorrectly called before the idr mechanism
is used to get the unit number, so the dd->unit hasn't been
initialized. This caused the unit relative directory creation to fail
after the first.

This patch moves the init for the debugfs stuff until after all of the
failures and after the unit number has been determined.

A bug in unwind code in qib_alloc_devdata() is also fixed.

Reviewed-by: Dennis Dalessandro <***@intel.com>
Signed-off-by: Mike Marciniszyn <***@intel.com>
Signed-off-by: Roland Dreier <***@purestorage.com>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
drivers/infiniband/hw/qib/qib_init.c | 25 +++++++++++--------------
1 file changed, 11 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c
index 24e802f4ea2f..76c3e177164d 100644
--- a/drivers/infiniband/hw/qib/qib_init.c
+++ b/drivers/infiniband/hw/qib/qib_init.c
@@ -1097,14 +1097,10 @@ struct qib_devdata *qib_alloc_devdata(struct pci_dev *pdev, size_t extra)
int ret;

dd = (struct qib_devdata *) ib_alloc_device(sizeof(*dd) + extra);
- if (!dd) {
- dd = ERR_PTR(-ENOMEM);
- goto bail;
- }
+ if (!dd)
+ return ERR_PTR(-ENOMEM);

-#ifdef CONFIG_DEBUG_FS
- qib_dbg_ibdev_init(&dd->verbs_dev);
-#endif
+ INIT_LIST_HEAD(&dd->list);

idr_preload(GFP_KERNEL);
spin_lock_irqsave(&qib_devs_lock, flags);
@@ -1121,11 +1117,6 @@ struct qib_devdata *qib_alloc_devdata(struct pci_dev *pdev, size_t extra)
if (ret < 0) {
qib_early_err(&pdev->dev,
"Could not allocate unit ID: error %d\n", -ret);
-#ifdef CONFIG_DEBUG_FS
- qib_dbg_ibdev_exit(&dd->verbs_dev);
-#endif
- ib_dealloc_device(&dd->verbs_dev.ibdev);
- dd = ERR_PTR(ret);
goto bail;
}

@@ -1139,9 +1130,15 @@ struct qib_devdata *qib_alloc_devdata(struct pci_dev *pdev, size_t extra)
qib_early_err(&pdev->dev,
"Could not alloc cpulist info, cpu affinity might be wrong\n");
}
-
-bail:
+#ifdef CONFIG_DEBUG_FS
+ qib_dbg_ibdev_init(&dd->verbs_dev);
+#endif
return dd;
+bail:
+ if (!list_empty(&dd->list))
+ list_del_init(&dd->list);
+ ib_dealloc_device(&dd->verbs_dev.ibdev);
+ return ERR_PTR(ret);;
}

/*
--
1.9.3
Jiri Slaby
2014-05-13 09:26:37 UTC
Permalink
From: David Fries <***@Fries.net>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit 6b355b33a64fd6d8ead2b838ec16fb9b551f71e8 upstream.

Previous logic,
if (avail > 8) {
store slave;
return;
}
send data; clear;

The logic error is, if there isn't space send the buffer and clear,
but the slave wasn't added to the now empty buffer loosing that slave
id. It also should have been "if (avail >= 8)" because when it is 8,
there is space.

Instead, if there isn't space send and clear the buffer, then there is
always space for the slave id.

Signed-off-by: David Fries <***@Fries.net>
Acked-by: Evgeniy Polyakov <***@ioremap.net>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
drivers/w1/w1_netlink.c | 25 +++++++++++++------------
1 file changed, 13 insertions(+), 12 deletions(-)

diff --git a/drivers/w1/w1_netlink.c b/drivers/w1/w1_netlink.c
index 40788c925d1c..73705aff53cb 100644
--- a/drivers/w1/w1_netlink.c
+++ b/drivers/w1/w1_netlink.c
@@ -54,28 +54,29 @@ static void w1_send_slave(struct w1_master *dev, u64 rn)
struct w1_netlink_msg *hdr = (struct w1_netlink_msg *)(msg + 1);
struct w1_netlink_cmd *cmd = (struct w1_netlink_cmd *)(hdr + 1);
int avail;
+ u64 *data;

/* update kernel slave list */
w1_slave_found(dev, rn);

avail = dev->priv_size - cmd->len;

- if (avail > 8) {
- u64 *data = (void *)(cmd + 1) + cmd->len;
+ if (avail < 8) {
+ msg->ack++;
+ cn_netlink_send(msg, 0, GFP_KERNEL);

- *data = rn;
- cmd->len += 8;
- hdr->len += 8;
- msg->len += 8;
- return;
+ msg->len = sizeof(struct w1_netlink_msg) +
+ sizeof(struct w1_netlink_cmd);
+ hdr->len = sizeof(struct w1_netlink_cmd);
+ cmd->len = 0;
}

- msg->ack++;
- cn_netlink_send(msg, 0, GFP_KERNEL);
+ data = (void *)(cmd + 1) + cmd->len;

- msg->len = sizeof(struct w1_netlink_msg) + sizeof(struct w1_netlink_cmd);
- hdr->len = sizeof(struct w1_netlink_cmd);
- cmd->len = 0;
+ *data = rn;
+ cmd->len += 8;
+ hdr->len += 8;
+ msg->len += 8;
}

static int w1_process_search_command(struct w1_master *dev, struct cn_msg *msg,
--
1.9.3
Jiri Slaby
2014-05-13 09:27:38 UTC
Permalink
From: Johannes Weiner <***@cmpxchg.org>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit 3a025760fc158b3726eac89ee95d7f29599e9dfa upstream.

On NUMA systems, a node may start thrashing cache or even swap anonymous
pages while there are still free pages on remote nodes.

This is a result of commits 81c0a2bb515f ("mm: page_alloc: fair zone
allocator policy") and fff4068cba48 ("mm: page_alloc: revert NUMA aspect
of fair allocation policy").

Before those changes, the allocator would first try all allowed zones,
including those on remote nodes, before waking any kswapds. But now,
the allocator fastpath doubles as the fairness pass, which in turn can
only consider the local node to prevent remote spilling based on
exhausted fairness batches alone. Remote nodes are only considered in
the slowpath, after the kswapds are woken up. But if remote nodes still
have free memory, kswapd should not be woken to rebalance the local node
or it may thrash cash or swap prematurely.

Fix this by adding one more unfair pass over the zonelist that is
allowed to spill to remote nodes after the local fairness pass fails but
before entering the slowpath and waking the kswapds.

This also gets rid of the GFP_THISNODE exemption from the fairness
protocol because the unfair pass is no longer tied to kswapd, which
GFP_THISNODE is not allowed to wake up.

However, because remote spills can be more frequent now - we prefer them
over local kswapd reclaim - the allocation batches on remote nodes could
underflow more heavily. When resetting the batches, use
atomic_long_read() directly instead of zone_page_state() to calculate the
delta as the latter filters negative counter values.

Signed-off-by: Johannes Weiner <***@cmpxchg.org>
Acked-by: Rik van Riel <***@redhat.com>
Acked-by: Mel Gorman <***@suse.de>
Signed-off-by: Andrew Morton <***@linux-foundation.org>
Signed-off-by: Linus Torvalds <***@linux-foundation.org>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
mm/internal.h | 1 +
mm/page_alloc.c | 89 +++++++++++++++++++++++++++++----------------------------
2 files changed, 46 insertions(+), 44 deletions(-)

diff --git a/mm/internal.h b/mm/internal.h
index 8b6cfd63b5a5..fdddbc83ac5f 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -369,5 +369,6 @@ unsigned long reclaim_clean_pages_from_list(struct zone *zone,
#define ALLOC_HIGH 0x20 /* __GFP_HIGH set */
#define ALLOC_CPUSET 0x40 /* check for correct cpuset */
#define ALLOC_CMA 0x80 /* allow allocations from CMA areas */
+#define ALLOC_FAIR 0x100 /* fair zone allocation */

#endif /* __MM_INTERNAL_H */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 6fca39097766..09459deb0b51 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1219,15 +1219,6 @@ void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
}
local_irq_restore(flags);
}
-static bool gfp_thisnode_allocation(gfp_t gfp_mask)
-{
- return (gfp_mask & GFP_THISNODE) == GFP_THISNODE;
-}
-#else
-static bool gfp_thisnode_allocation(gfp_t gfp_mask)
-{
- return false;
-}
#endif

/*
@@ -1564,12 +1555,7 @@ again:
get_pageblock_migratetype(page));
}

- /*
- * NOTE: GFP_THISNODE allocations do not partake in the kswapd
- * aging protocol, so they can't be fair.
- */
- if (!gfp_thisnode_allocation(gfp_flags))
- __mod_zone_page_state(zone, NR_ALLOC_BATCH, -(1 << order));
+ __mod_zone_page_state(zone, NR_ALLOC_BATCH, -(1 << order));

__count_zone_vm_events(PGALLOC, zone, 1 << order);
zone_statistics(preferred_zone, zone, gfp_flags);
@@ -1935,23 +1921,12 @@ zonelist_scan:
* zone size to ensure fair page aging. The zone a
* page was allocated in should have no effect on the
* time the page has in memory before being reclaimed.
- *
- * Try to stay in local zones in the fastpath. If
- * that fails, the slowpath is entered, which will do
- * another pass starting with the local zones, but
- * ultimately fall back to remote zones that do not
- * partake in the fairness round-robin cycle of this
- * zonelist.
- *
- * NOTE: GFP_THISNODE allocations do not partake in
- * the kswapd aging protocol, so they can't be fair.
*/
- if ((alloc_flags & ALLOC_WMARK_LOW) &&
- !gfp_thisnode_allocation(gfp_mask)) {
- if (zone_page_state(zone, NR_ALLOC_BATCH) <= 0)
- continue;
+ if (alloc_flags & ALLOC_FAIR) {
if (!zone_local(preferred_zone, zone))
continue;
+ if (zone_page_state(zone, NR_ALLOC_BATCH) <= 0)
+ continue;
}
/*
* When allocating a page cache page for writing, we
@@ -2399,32 +2374,40 @@ __alloc_pages_high_priority(gfp_t gfp_mask, unsigned int order,
return page;
}

-static void prepare_slowpath(gfp_t gfp_mask, unsigned int order,
- struct zonelist *zonelist,
- enum zone_type high_zoneidx,
- struct zone *preferred_zone)
+static void reset_alloc_batches(struct zonelist *zonelist,
+ enum zone_type high_zoneidx,
+ struct zone *preferred_zone)
{
struct zoneref *z;
struct zone *zone;

for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
- if (!(gfp_mask & __GFP_NO_KSWAPD))
- wakeup_kswapd(zone, order, zone_idx(preferred_zone));
/*
* Only reset the batches of zones that were actually
- * considered in the fast path, we don't want to
- * thrash fairness information for zones that are not
+ * considered in the fairness pass, we don't want to
+ * trash fairness information for zones that are not
* actually part of this zonelist's round-robin cycle.
*/
if (!zone_local(preferred_zone, zone))
continue;
mod_zone_page_state(zone, NR_ALLOC_BATCH,
- high_wmark_pages(zone) -
- low_wmark_pages(zone) -
- zone_page_state(zone, NR_ALLOC_BATCH));
+ high_wmark_pages(zone) - low_wmark_pages(zone) -
+ atomic_long_read(&zone->vm_stat[NR_ALLOC_BATCH]));
}
}

+static void wake_all_kswapds(unsigned int order,
+ struct zonelist *zonelist,
+ enum zone_type high_zoneidx,
+ struct zone *preferred_zone)
+{
+ struct zoneref *z;
+ struct zone *zone;
+
+ for_each_zone_zonelist(zone, z, zonelist, high_zoneidx)
+ wakeup_kswapd(zone, order, zone_idx(preferred_zone));
+}
+
static inline int
gfp_to_alloc_flags(gfp_t gfp_mask)
{
@@ -2513,12 +2496,13 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
* allowed per node queues are empty and that nodes are
* over allocated.
*/
- if (gfp_thisnode_allocation(gfp_mask))
+ if (IS_ENABLED(CONFIG_NUMA) &&
+ (gfp_mask & GFP_THISNODE) == GFP_THISNODE)
goto nopage;

restart:
- prepare_slowpath(gfp_mask, order, zonelist,
- high_zoneidx, preferred_zone);
+ if (!(gfp_mask & __GFP_NO_KSWAPD))
+ wake_all_kswapds(order, zonelist, high_zoneidx, preferred_zone);

/*
* OK, we're below the kswapd watermark and have kicked background
@@ -2695,7 +2679,7 @@ __alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order,
struct page *page = NULL;
int migratetype = allocflags_to_migratetype(gfp_mask);
unsigned int cpuset_mems_cookie;
- int alloc_flags = ALLOC_WMARK_LOW|ALLOC_CPUSET;
+ int alloc_flags = ALLOC_WMARK_LOW|ALLOC_CPUSET|ALLOC_FAIR;
struct mem_cgroup *memcg = NULL;

gfp_mask &= gfp_allowed_mask;
@@ -2736,12 +2720,29 @@ retry_cpuset:
if (allocflags_to_migratetype(gfp_mask) == MIGRATE_MOVABLE)
alloc_flags |= ALLOC_CMA;
#endif
+retry:
/* First allocation attempt */
page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, nodemask, order,
zonelist, high_zoneidx, alloc_flags,
preferred_zone, migratetype);
if (unlikely(!page)) {
/*
+ * The first pass makes sure allocations are spread
+ * fairly within the local node. However, the local
+ * node might have free pages left after the fairness
+ * batches are exhausted, and remote zones haven't
+ * even been considered yet. Try once more without
+ * fairness, and include remote zones now, before
+ * entering the slowpath and waking kswapd: prefer
+ * spilling to a remote zone over swapping locally.
+ */
+ if (alloc_flags & ALLOC_FAIR) {
+ reset_alloc_batches(zonelist, high_zoneidx,
+ preferred_zone);
+ alloc_flags &= ~ALLOC_FAIR;
+ goto retry;
+ }
+ /*
* Runtime PM, block IO and its error handling path
* can deadlock because I/O on the device might not
* complete.
--
1.9.3
Jiri Slaby
2014-05-13 09:28:04 UTC
Permalink
From: Huacai Chen <***@lemote.com>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit c14af233fbe279d0e561ecf84f1208b1bae087ef upstream.

The original MIPS hibernate code flushes cache and TLB entries in
swsusp_arch_resume(). But they are removed in Commit 44eeab67416711
(MIPS: Hibernation: Remove SMP TLB and cacheflushing code.). A cross-
CPU flush is surely unnecessary because all but the local CPU have
already been disabled. But a local flush (at least the TLB flush) is
needed. When we do hibernation on Loongson-3 with an E1000E NIC, it is
very easy to produce a kernel panic (kernel page fault, or unaligned
access). The root cause is E1000E driver use vzalloc_node() to allocate
pages, the stale TLB entries of the booting kernel will be misused by
the resumed target kernel.

Signed-off-by: Huacai Chen <***@lemote.com>
Cc: John Crispin <***@phrozen.org>
Cc: Steven J. Hill <***@imgtec.com>
Cc: Aurelien Jarno <***@aurel32.net>
Cc: linux-***@linux-mips.org
Cc: Fuxin Zhang <***@lemote.com>
Cc: Zhangjin Wu <***@gmail.com>
Patchwork: https://patchwork.linux-mips.org/patch/6643/
Signed-off-by: Ralf Baechle <***@linux-mips.org>
Signed-off-by: Jiri Slaby <***@suse.cz>
---
arch/mips/power/hibernate.S | 1 +
1 file changed, 1 insertion(+)

diff --git a/arch/mips/power/hibernate.S b/arch/mips/power/hibernate.S
index 7e0277a1048f..32a7c828f073 100644
--- a/arch/mips/power/hibernate.S
+++ b/arch/mips/power/hibernate.S
@@ -43,6 +43,7 @@ LEAF(swsusp_arch_resume)
bne t1, t3, 1b
PTR_L t0, PBE_NEXT(t0)
bnez t0, 0b
+ jal local_flush_tlb_all /* Avoid TLB mismatch after kernel resume */
PTR_LA t0, saved_regs
PTR_L ra, PT_R31(t0)
PTR_L sp, PT_R29(t0)
--
1.9.3
Jiri Slaby
2014-05-13 09:28:42 UTC
Permalink
From: Giacomo Comes <***@naic.edu>