Commit graph

361,091 commits

Author SHA1 Message Date
Vasanthakumar Thiagarajan
9d9188409a ath6kl: Fix bug in scheduling hb_timer
hb_timer should be scheduled only when hb_poll is non-zero.
But in ath6kl_recovery_work() the timer is scheduled based
on fw_recovery.enable instead which is wrong.

Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:49 +03:00
Vasanthakumar Thiagarajan
77565794eb ath6kl: Recover from "wmi ctrl ep is full" condition
In some error conditions, fw pauses HTC pipes which would
result in control endpoint full condition. When we hit this
case, most of the time the device will be unusable. Re-initialize
the target to recover from this situation.

Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:49 +03:00
Vasanthakumar Thiagarajan
9233299394 ath6kl: Add support to detect fw error through heart beat
This patch adds support to detect fw error condition by sending
periodic message (heart beat challenge) to firmware. Upon reception
of the message, fw would send a response event to driver. When
there are no reponses from fw for about 5 cmd driver would
trigger the recovery logic assuming that fw has gone into an
error state.

Capable fw will advertise this capability through
ATH6KL_FW_CAPABILITY_HEART_BEAT_POLL bit. This feature
is disabled by default, can be enabled through a modparam
(heart_beat_poll). This modparam also confiures the polling
interval in msecs.

Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:48 +03:00
Vasanthakumar Thiagarajan
84caf8005b ath6kl: Recover from fw crash
Re-initialize the target when fw crash is reported.
This would make the device functional again after
target crash. During the target re-initialization
it is made sure that target is not bugged with data/cmd
request, ar->state ATH6KL_STATE_RECOVERY is used
for this purpose.

Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:48 +03:00
Vasanthakumar Thiagarajan
ede615d2f0 ath6kl: Refactor ath6kl_init_hw_start() and ath6kl_init_hw_stop()
So that these functions will be used to re-initialize the fw
upon detecting fw error. This refactoring moves ar->state
setting out of core stop/start functionality.

Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:48 +03:00
Dengke Qiu
83685091ac ath6kl: fix link speed when using sgi
The MSB of rate index from FW is used for sgi. But the ath6kl_wmi_get_rate
doesn't handle it. The access to wmi_rate_tbl array may be out of range
if sgi is 1. This may cause the return value of ath6kl_wmi_get_rate()
function is incorrect link rate. We add sgi adjustment to avoid such case.

kvalo: change patch title

Signed-off-by: Dengke Qiu <dqiu@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:48 +03:00
Mohammed Shafi Shajakhan
3814264481 ath6kl: trivial cleanup on interface type selection
a minor cleanup in assigning the driver specific network type
based on interface type.

Signed-off-by: Mohammed Shafi Shajakhan <mohammed@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:48 +03:00
Thomas Pedersen
2c07cf4461 ath6kl: consolidate WoW pattern length
Since WOW_MASK_SIZE and WOW_PATTERN_SIZE have the same value, are
logically equivalent, and part of the WMI API so therefore unlikely to
change, consolidate these into WOW_PATTERN_SIZE.

Reported-by Kalle Valo <kvalo@qualcomm.com>
Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:47 +03:00
Thomas Pedersen
b1f47e3a96 ath6kl: rework scheduled scan
This patch reflects changes in the firmware scheduled scan
implementation to behave better in cases with multiple concurrent vifs.
Major changes:

	- scheduled scan filters and state are now programmed per-vif.
	- decouple scheduled scan from host sleep.

To maintain graceful failure with old firmwares, a new firmware
capability bit is introduced: ATH6KL_FW_CAPABILITY_SCHED_SCAN_V2.
ath6kl simply won't advertise scheduled scan to cfg80211 if the
SCHED_SCAN_V2 is not supported.

Since firmwares from here on out won't support the previous implicit API
for scheduled scan (set WoW filters and host sleep), bump the firmware
API to protect old drivers.

Unfortunately, due to firmware RAM constraints ath6kl still cannot
expect a scan complete event at the end of a scheduled scan results
cycle, so the sched_scan_timer is retained.

Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:46 +03:00
Marina Makienko
a3b3842c2e ath6kl: check usb_register() return value
ath6kl_usb_init() does not check usb_register() return value.
As a result it may incorrectly report success of driver initialization.

Found by Linux Driver Verification project (linuxtesting.org).

kvalo: fix commit title and make cosmetic changes to the code to follow
more the style used in the driver

Signed-off-by: Marina Makienko <makienko@ispras.ru>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:46 +03:00
Vasanthakumar Thiagarajan
8114f9b6d2 ath6kl: Fix potential memory leak in ath6kl_tx_complete()
We bail out from ath6kl_tx_complete() if any of the sanity
checks on skb and ath6kl_cookie fails. By doing this we
potentially leak few remaining buffers in packet_queue.
Make sure to proceed processing the remaining buffers
as well. This issue is found during code review.

Reported-by: Wang yufeng <yufengw@qca.qualcomm.com>
Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:45 +03:00
Vasanthakumar Thiagarajan
0616dc1f2b ath6kl: Fix potential skb double free in ath6kl_wmi_sync_point()
skb given to ath6kl_control_tx() is owned by ath6kl_control_tx().
Calling function should not free the skb for error cases.
This is found during code review.

kvalo: fix a checkpatch warning in ath6kl_wmi_cmd_send()

Signed-off-by: Vasanthakumar Thiagarajan <vthiagar@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:45 +03:00
Thomas Pedersen
f21243a822 ath6kl: reconfigure RSN capabilities when restarting AP
If the firmware decides to initiate a channel switch on an AP vif
running an RSN BSS, reconfigure the saved RSN IE capabilities as well.

Fixes a bug where the beacon and 4-way handshake would have a capability
mismatch after a channel switch, since the firmware apparently clears
these on an AP disconnect.

Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:42 +03:00
Thomas Pedersen
b5495e666d ath6kl: restart concurrent vifs on failed connect
When an ath6kl STA vif is issued a connect command, the firmware will
disconnect all other beaconing vifs in preparation for a potential
channel switch. The case where the connect fails is currently unhandled,
so if a connection attempt on a STA vif fails and any vifs were waiting
for a new channel, simply restart the concurrent vifs on their previous
channel.

Requires that we start tracking the last issued channel in ar->last_ch,
which is valid since ath6kl only supports 1 channel at a time.

Also clear the beaconing vif's want_ch_switch bit regardless of whether
channel switch succeeds, to stop recommitting the same failed profile.

Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:40 +03:00
Thomas Pedersen
fd4377b6ba ath6kl: configure wow filters per-vif
Only WoW filters for the first vif were being set, causing failures to
wake up on any concurrent connected vifs. Handle all per-vif suspend
and resume tasks.

Since cfg80211 issues user wow filters on a per-wiphy basis, set any
custom filters on all connected vifs.

Starting WoW in firmware and setting host sleep mode is still handled on
a global per-phy level. The first vif is always used for bookkeeping
regardless of whether it is connected or not.

WoW is cancelled if no connected vifs are found.

No firmware capability bits or API bump is needed for this patch, as
setting filters for vifs with index > 0 will simply overwrite the index
0 filters in the current implementation. While not correct, this is
identical to the existing behavior.

kvalo: fix a checkpatch warning in ath6kl_wow_resume()

Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:39 +03:00
Kalle Valo
f8c0305383 ath6kl: fix incorrect use of IEEE80211_NUM_BANDS
ath6kl was incorrectly assuming that IEEE80211_NUM_BANDS will always be 2
and used that also in the firmware WMI interface definitions. But after
the support for 60 GHz was added to cfg80211 IEEE80211_NUM_BANDS changed to 3
and this can cause all sort of problems, possibly even memory corruption.
I only found this during code review and didn't notice any bugs, but I'm
sure there are a few lurking somewhere.

To fix this rename unused A_NUM_BANDS to ATH6KL_NUM_BANDS, which is
always defined to be 2, and use that in WMI.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:37 +03:00
Kalle Valo
84841ba29b ath6kl: add support for changing contry code
To make it possible to change the country code from user space via nl80211
add handler for reg_notifier. The feature is only enabled when built
time option CONFIG_ATH6KL_REGDOMAIN is enabled, which again depends on
CFG80211_CERTIFICATION_ONUS for certication purposes.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:37 +03:00
Kalle Valo
11f0bfcf73 ath6kl: refactor wmi scan command
ATH6KL_FW_CAPABILITY_STA_P2PDEV_DUPLEX was checked in cfg80211.c which is
a bit awkward when adding more callsites to the scan functions. Refactor
the code to wmi.c so that it's transparent to the callers.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:35 +03:00
Kalle Valo
c8c72b74e2 ath6kl: move ath6kl_wmi_startscan_cmd()
To make it easier to refactor the scan commands move
ath6kl_wmi_startscan_cmd() before the beginscan function. No functional
changes.

Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:34 +03:00
Thomas Pedersen
279b2862ee ath6kl: support TX error rate notification
The ath6kl firmware can monitor a connection and report when a certain
TX failure threshold is crossed. Support this configuration and event
reporting on compatible firmwares.

Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:33 +03:00
Bala Shanmugam
bf744f1178 ath6kl: Add support for AR6004 hardware version 1.3
Add support for AR6004 hardware with version 1.3 and has
id 0x31c8088a.

Signed-off-by: Bala Shanmugam <bkamatch@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:32 +03:00
Pandiyarajan Pitchaimuthu
4aca81bfb0 ath6kl: Make use of return value from ath6kl_diag_read()
In ath6kl_read_fwlogs(), return value from ath6kl_diag_read()is not
used to bail out in case of any errors in reading fw log. No real issue
is observed because of this, reported by source code analyzer.

kvalo: fix a long line warning

Signed-off-by: Pandiyarajan Pitchaimuthu <c_ppitch@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:32 +03:00
Aarthi Thiruvengadam
c95dcb595d ath6kl: use custom MAC address for newly created interfaces
Firmware and driver generate MAC addresses for the second and third interfaces.
In addition to the existing algorithm, flip bit 7 of 5th octet. Since both
firmware and driver individually generate the MAC addresses, introduce a new
firmware capability bit to keep them compatible.

Signed-off-by: Aarthi Thiruvengadam <athiruve@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:32 +03:00
Thomas Pedersen
85b20fc242 ath6kl: support rssi threshold for sched scan
The ath6kl firmware can filter scan results based on rssi. This is
useful to limit hosts wakeups on scheduled scans.

Signed-off-by: Thomas Pedersen <c_tpeder@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
2012-10-24 11:49:31 +03:00
Jiri Olsa
3f3a206487 perf test: Add automated tests for pmu sysfs translated events
Add automated tests for all events found under PMU/events
directory. Tested events are in the 'cpu/event=xxx/u' format,
where 'xxx' is substituted by every event found.

The 'event=xxx' term is translated to the cpu specific term.
We only check that the event is created (not the real config
numbers) and that the modifier is properly set.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-9-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:27 +02:00
Jiri Olsa
1d33d6dce1 perf tools: Add support to specify hw event as PMU event term
Add a way to specify hw event as PMU event term like:

 'cpu/event=cpu-cycles/u'
 'cpu/event=instructions,.../u'
 'cpu/cycles,.../u'

The 'event=cpu-cycles' term is replaced/translated by the hw events
term translation, which is exposed by sysfs 'events' group attribute.

Add parser bits, the rest is already handled by the PMU alias code.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-8-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:27 +02:00
Jiri Olsa
3fded963cd perf tools: Fix PMU object alias initialization
The pmu_lookup should return pmus that do not expose the 'events'
group attribute in sysfs. Also it should fail when any other error
during 'events' lookup is hit (pmu_aliases fails).

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-7-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:26 +02:00
Jiri Olsa
20550a4345 perf/x86: Add hardware events translations for Intel P6 cpus
Add support for Intel P6 processors to display 'events' sysfs
directory (/sys/devices/cpu/events/) with hw event translations:

  # ls /sys/devices/cpu/events/
  branch-instructions
  branch-misses
  bus-cycles
  cache-misses
  cache-references
  cpu-cycles
  instructions
  ref-cycles
  stalled-cycles-backend
  stalled-cycles-frontend

Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-6-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:25 +02:00
Jiri Olsa
0bf79d4413 perf/x86: Add hardware events translations for AMD cpus
Add support for AMD processors to display 'events' sysfs
directory (/sys/devices/cpu/events/) with hw event translations:

  # ls  /sys/devices/cpu/events/
  branch-instructions
  branch-misses
  bus-cycles
  cache-misses
  cache-references
  cpu-cycles
  instructions
  ref-cycles
  stalled-cycles-backend
  stalled-cycles-frontend

Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-5-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:25 +02:00
Jiri Olsa
43c032febd perf/x86: Add hardware events translations for Intel cpus
Add support for Intel processors to display 'events' sysfs
directory (/sys/devices/cpu/events/) with hw event translations:

  # ls  /sys/devices/cpu/events/
  branch-instructions
  branch-misses
  bus-cycles
  cache-misses
  cache-references
  cpu-cycles
  instructions
  ref-cycles
  stalled-cycles-backend
  stalled-cycles-frontend

Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-4-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:24 +02:00
Jiri Olsa
8300daa267 perf/x86: Filter out undefined events from sysfs events attribute
The sysfs events group attribute currently shows all hw events,
including also undefined ones.

This patch filters out all undefined events out of the sysfs events
group attribute, so they don't even show up.

Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-3-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:24 +02:00
Jiri Olsa
a47473939d perf/x86: Make hardware event translations available in sysfs
Add support to display hardware events translations available
through the sysfs. Add 'events' group attribute under the sysfs
x86 PMU record with attribute/file for each hardware event.

This patch adds only backbone for PMUs to display config under
'events' directory. The specific PMU support itself will come
in next patches, however this is how the sysfs group will look
like:

  # ls  /sys/devices/cpu/events/
  branch-instructions
  branch-misses
  bus-cycles
  cache-misses
  cache-references
  cpu-cycles
  instructions
  ref-cycles
  stalled-cycles-backend
  stalled-cycles-frontend

The file - hw event ID mapping is:

  file                      hw event ID
  ---------------------------------------------------------------
  cpu-cycles                PERF_COUNT_HW_CPU_CYCLES
  instructions              PERF_COUNT_HW_INSTRUCTIONS
  cache-references          PERF_COUNT_HW_CACHE_REFERENCES
  cache-misses              PERF_COUNT_HW_CACHE_MISSES
  branch-instructions       PERF_COUNT_HW_BRANCH_INSTRUCTIONS
  branch-misses             PERF_COUNT_HW_BRANCH_MISSES
  bus-cycles                PERF_COUNT_HW_BUS_CYCLES
  stalled-cycles-frontend   PERF_COUNT_HW_STALLED_CYCLES_FRONTEND
  stalled-cycles-backend    PERF_COUNT_HW_STALLED_CYCLES_BACKEND
  ref-cycles                PERF_COUNT_HW_REF_CPU_CYCLES

Each file in the 'events' directory contains the term translation
for the symbolic hw event for the currently running cpu model.

  # cat /sys/devices/cpu/events/stalled-cycles-backend
  event=0xb1,umask=0x01,inv,cmask=0x01

Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1349873598-12583-2-git-send-email-jolsa@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:41:23 +02:00
Vince Weaver
58e9eaf06f perf/x86: Remove P6 cpuc->enabled check
Between 2.6.33 and 2.6.34 the PMU code was made modular.

The x86_pmu_enable() call was extended to disable cpuc->enabled
and iterate the counters, enabling one at a time, before calling
enable_all() at the end, followed by re-enabling cpuc->enabled.

Since cpuc->enabled was set to 0, that change effectively caused
the "val |= ARCH_PERFMON_EVENTSEL_ENABLE;" code in p6_pmu_enable_event()
and p6_pmu_disable_event() to be dead code that was never called.

This change removes this code (which was confusing) and adds some
extra commentary to make it more clear what is going on.

Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1210191732000.14552@vincent-weaver-1.um.maine.edu
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:32:00 +02:00
Vince Weaver
e09df47885 perf/x86: Update/fix generic events on P6 PMU
This patch updates the generic events on p6, including some new
extended cache events.

Values for these events were taken from the equivelant PAPI
predefined events.

Tested on a Pentium II.

Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1210191730080.14552@vincent-weaver-1.um.maine.edu
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:31:58 +02:00
Vince Weaver
7991c9ca40 perf/x86: Fix P6 FP_ASSIST event constraint
According to Intel SDM Volume 3B, FP_ASSIST is limited to Counter 1 only,
not Counter 0.

Tested on a Pentium II.

Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/alpine.DEB.2.02.1210191728570.14552@vincent-weaver-1.um.maine.edu
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:31:57 +02:00
Peter Zijlstra
e9c84cb8d5 sched: Describe CFS load-balancer
Add some scribbles on how and why the load-balancer works..

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1341316406.23484.64.camel@twins
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:33 +02:00
Paul Turner
f4e26b120b sched: Introduce temporary FAIR_GROUP_SCHED dependency for load-tracking
While per-entity load-tracking is generally useful, beyond computing shares
distribution, e.g. runnable based load-balance (in progress), governors,
power-management, etc.

These facilities are not yet consumers of this data.  This may be trivially
reverted when the information is required; but avoid paying the overhead for
calculations we will not use until then.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141507.422162369@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:31 +02:00
Paul Turner
5b51f2f80b sched: Make __update_entity_runnable_avg() fast
__update_entity_runnable_avg forms the core of maintaining an entity's runnable
load average.  In this function we charge the accumulated run-time since last
update and handle appropriate decay.  In some cases, e.g. a waking task, this
time interval may be much larger than our period unit.

Fortunately we can exploit some properties of our series to perform decay for a
blocked update in constant time and account the contribution for a running
update in essentially-constant* time.

[*]: For any running entity they should be performing updates at the tick which
gives us a soft limit of 1 jiffy between updates, and we can compute up to a
32 jiffy update in a single pass.

C program to generate the magic constants in the arrays:

  #include <math.h>
  #include <stdio.h>

  #define N 32
  #define WMULT_SHIFT 32

  const long WMULT_CONST = ((1UL << N) - 1);
  double y;

  long runnable_avg_yN_inv[N];
  void calc_mult_inv() {
  	int i;
  	double yn = 0;

  	printf("inverses\n");
  	for (i = 0; i < N; i++) {
  		yn = (double)WMULT_CONST * pow(y, i);
  		runnable_avg_yN_inv[i] = yn;
  		printf("%2d: 0x%8lx\n", i, runnable_avg_yN_inv[i]);
  	}
  	printf("\n");
  }

  long mult_inv(long c, int n) {
  	return (c * runnable_avg_yN_inv[n]) >>  WMULT_SHIFT;
  }

  void calc_yn_sum(int n)
  {
  	int i;
  	double sum = 0, sum_fl = 0, diff = 0;

  	/*
  	 * We take the floored sum to ensure the sum of partial sums is never
  	 * larger than the actual sum.
  	 */
  	printf("sum y^n\n");
  	printf("   %8s  %8s %8s\n", "exact", "floor", "error");
  	for (i = 1; i <= n; i++) {
  		sum = (y * sum + y * 1024);
  		sum_fl = floor(y * sum_fl+ y * 1024);
  		printf("%2d: %8.0f  %8.0f %8.0f\n", i, sum, sum_fl,
  			sum_fl - sum);
  	}
  	printf("\n");
  }

  void calc_conv(long n) {
  	long old_n;
  	int i = -1;

  	printf("convergence (LOAD_AVG_MAX, LOAD_AVG_MAX_N)\n");
  	do {
  		old_n = n;
  		n = mult_inv(n, 1) + 1024;
  		i++;
  	} while (n != old_n);
  	printf("%d> %ld\n", i - 1, n);
  	printf("\n");
  }

  void main() {
  	y = pow(0.5, 1/(double)N);
  	calc_mult_inv();
  	calc_conv(1024);
  	calc_yn_sum(N);
  }

[ Compile with -lm ]
Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141507.277808946@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:30 +02:00
Paul Turner
f269ae0469 sched: Update_cfs_shares at period edge
Now that our measurement intervals are small (~1ms) we can amortize the posting
of update_shares() to be about each period overflow.  This is a large cost
saving for frequently switching tasks.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141507.200772172@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:29 +02:00
Paul Turner
48a1675323 sched: Refactor update_shares_cpu() -> update_blocked_avgs()
Now that running entities maintain their own load-averages the work we must do
in update_shares() is largely restricted to the periodic decay of blocked
entities.  This allows us to be a little less pessimistic regarding our
occupancy on rq->lock and the associated rq->clock updates required.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141507.133999170@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:28 +02:00
Paul Turner
82958366cf sched: Replace update_shares weight distribution with per-entity computation
Now that the machinery in place is in place to compute contributed load in a
bottom up fashion; replace the shares distribution code within update_shares()
accordingly.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141507.061208672@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:28 +02:00
Paul Turner
f1b17280ef sched: Maintain runnable averages across throttled periods
With bandwidth control tracked entities may cease execution according to user
specified bandwidth limits.  Charging this time as either throttled or blocked
however, is incorrect and would falsely skew in either direction.

What we actually want is for any throttled periods to be "invisible" to
load-tracking as they are removed from the system for that interval and
contribute normally otherwise.

Do this by moderating the progression of time to omit any periods in which the
entity belonged to a throttled hierarchy.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.998912151@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:27 +02:00
Paul Turner
bb17f65571 sched: Normalize tg load contributions against runnable time
Entities of equal weight should receive equitable distribution of cpu time.
This is challenging in the case of a task_group's shares as execution may be
occurring on multiple cpus simultaneously.

To handle this we divide up the shares into weights proportionate with the load
on each cfs_rq.  This does not however, account for the fact that the sum of
the parts may be less than one cpu and so we need to normalize:
  load(tg) = min(runnable_avg(tg), 1) * tg->shares
Where runnable_avg is the aggregate time in which the task_group had runnable
children.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.930124292@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:26 +02:00
Paul Turner
8165e145ce sched: Compute load contribution by a group entity
Unlike task entities who have a fixed weight, group entities instead own a
fraction of their parenting task_group's shares as their contributed weight.

Compute this fraction so that we can correctly account hierarchies and shared
entity nodes.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.855074415@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:25 +02:00
Paul Turner
c566e8e9e4 sched: Aggregate total task_group load
Maintain a global running sum of the average load seen on each cfs_rq belonging
to each task group so that it may be used in calculating an appropriate
shares:weight distribution.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.792901086@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:24 +02:00
Paul Turner
aff3e49884 sched: Account for blocked load waking back up
When a running entity blocks we migrate its tracked load to
cfs_rq->blocked_runnable_avg.  In the sleep case this occurs while holding
rq->lock and so is a natural transition.  Wake-ups however, are potentially
asynchronous in the presence of migration and so special care must be taken.

We use an atomic counter to track such migrated load, taking care to match this
with the previously introduced decay counters so that we don't migrate too much
load.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.726077467@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:23 +02:00
Paul Turner
0a74bef8be sched: Add an rq migration call-back to sched_class
Since we are now doing bottom up load accumulation we need explicit
notification when a task has been re-parented so that the old hierarchy can be
updated.

Adds: migrate_task_rq(struct task_struct *p, int next_cpu)

(The alternative is to do this out of __set_task_cpu, but it was suggested that
this would be a cleaner encapsulation.)

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.660023400@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:23 +02:00
Paul Turner
9ee474f556 sched: Maintain the load contribution of blocked entities
We are currently maintaining:

  runnable_load(cfs_rq) = \Sum task_load(t)

For all running children t of cfs_rq.  While this can be naturally updated for
tasks in a runnable state (as they are scheduled); this does not account for
the load contributed by blocked task entities.

This can be solved by introducing a separate accounting for blocked load:

  blocked_load(cfs_rq) = \Sum runnable(b) * weight(b)

Obviously we do not want to iterate over all blocked entities to account for
their decay, we instead observe that:

  runnable_load(t) = \Sum p_i*y^i

and that to account for an additional idle period we only need to compute:

  y*runnable_load(t).

This means that we can compute all blocked entities at once by evaluating:

  blocked_load(cfs_rq)` = y * blocked_load(cfs_rq)

Finally we maintain a decay counter so that when a sleeping entity re-awakens
we can determine how much of its load should be removed from the blocked sum.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.585389902@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:22 +02:00
Paul Turner
2dac754e10 sched: Aggregate load contributed by task entities on parenting cfs_rq
For a given task t, we can compute its contribution to load as:

  task_load(t) = runnable_avg(t) * weight(t)

On a parenting cfs_rq we can then aggregate:

  runnable_load(cfs_rq) = \Sum task_load(t), for all runnable children t

Maintain this bottom up, with task entities adding their contributed load to
the parenting cfs_rq sum.  When a task entity's load changes we add the same
delta to the maintained sum.

Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.514678907@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:21 +02:00
Ben Segall
18bf2805d9 sched: Maintain per-rq runnable averages
Since runqueues do not have a corresponding sched_entity we instead embed a
sched_avg structure directly.

Signed-off-by: Ben Segall <bsegall@google.com>
Reviewed-by: Paul Turner <pjt@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141506.442637130@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2012-10-24 10:27:20 +02:00