Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1397324
| From | Kamal Mostafa <kamal@canonical.com> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 4.2.y-ckt 22/59] drm/i915: Make RPS EI/thresholds multiple of 25 on SNB-BDW |
| Date | 2016-05-09 22:10 +0200 |
| Message-ID | <rwZFa-42D-57@gated-at.bofh.it> (permalink) |
| References | <rwZvs-3Ct-3@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
4.2.8-ckt10 -stable review patch. If anyone has any objections, please let me know.
---8<------------------------------------------------------------
From: =?UTF-8?q?Ville=20Syrj=C3=A4l=C3=A4?= <ville.syrjala@linux.intel.com>
commit 4ea3959018d09edfa36a9e7b5ccdbd4ec4b99e49 upstream.
Somehow my SNB GT1 (Dell XPS 8300) gets very unhappy around
GPU hangs if the RPS EI/thresholds aren't suitably aligned.
It seems like scheduling/timer interupts stop working somehow
and things get stuck eg. in usleep_range().
I bisected the problem down to
commit 8a5864377b12 ("drm/i915/skl: Restructured the gen6_set_rps_thresholds function")
I observed that before all the values were at least multiples of 25,
but afterwards they are not. And rounding things up to the next multiple
of 25 does seem to help, so lets' do that. I also tried roundup(..., 5)
but that wasn't sufficient. Also I have no idea if we might need this sort of
thing on gen9+ as well.
These are the original EI/thresholds:
LOW_POWER
GEN6_RP_UP_EI 12500
GEN6_RP_UP_THRESHOLD 11800
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 21250
BETWEEN
GEN6_RP_UP_EI 10250
GEN6_RP_UP_THRESHOLD 9225
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 18750
HIGH_POWER
GEN6_RP_UP_EI 8000
GEN6_RP_UP_THRESHOLD 6800
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 15000
These are after 8a5864377b12:
LOW_POWER
GEN6_RP_UP_EI 12500
GEN6_RP_UP_THRESHOLD 11875
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 21250
BETWEEN
GEN6_RP_UP_EI 10156
GEN6_RP_UP_THRESHOLD 9140
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 18750
HIGH_POWER
GEN6_RP_UP_EI 7812
GEN6_RP_UP_THRESHOLD 6640
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 15000
And these are what we have after this patch:
LOW_POWER
GEN6_RP_UP_EI 12500
GEN6_RP_UP_THRESHOLD 11875
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 21250
BETWEEN
GEN6_RP_UP_EI 10175
GEN6_RP_UP_THRESHOLD 9150
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 18750
HIGH_POWER
GEN6_RP_UP_EI 7825
GEN6_RP_UP_THRESHOLD 6650
GEN6_RP_DOWN_EI 25000
GEN6_RP_DOWN_THRESHOLD 15000
Cc: Akash Goel <akash.goel@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Testcase: igt/kms_pipe_crc_basic/hang-read-crc-pipe-B
Fixes: 8a5864377b12 ("drm/i915/skl: Restructured the gen6_set_rps_thresholds function")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1461159836-9108-1-git-send-email-ville.syrjala@linux.intel.com
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
(cherry picked from commit 8a292d016d1cc4938ff14b4df25328230b08a408)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Kamal Mostafa <kamal@canonical.com>
---
drivers/gpu/drm/i915/i915_reg.h | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index 2030f60..3888d13 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -2721,7 +2721,14 @@ enum skl_disp_power_wells {
#define GEN6_RP_STATE_LIMITS (MCHBAR_MIRROR_BASE_SNB + 0x5994)
#define GEN6_RP_STATE_CAP (MCHBAR_MIRROR_BASE_SNB + 0x5998)
-#define INTERVAL_1_28_US(us) (((us) * 100) >> 7)
+/*
+ * Make these a multiple of magic 25 to avoid SNB (eg. Dell XPS
+ * 8300) freezing up around GPU hangs. Looks as if even
+ * scheduling/timer interrupts start misbehaving if the RPS
+ * EI/thresholds are "bad", leading to a very sluggish or even
+ * frozen machine.
+ */
+#define INTERVAL_1_28_US(us) roundup(((us) * 100) >> 7, 25)
#define INTERVAL_1_33_US(us) (((us) * 3) >> 2)
#define GT_INTERVAL_FROM_US(dev_priv, us) (IS_GEN9(dev_priv) ? \
INTERVAL_1_33_US(us) : \
--
2.7.4
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[4.2.y-ckt stable] Linux 4.2.8-ckt10 stable review Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 25/59] ARM: SoCFPGA: Fix secondary CPU startup in thumb2 kernel Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 57/59] net/mlx5e: Fix MLX5E_100BASE_T define Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 53/59] batman-adv: Check skb size before using encapsulated ETH+VLAN header Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 04/59] ASoC: dapm: Make sure we have a card when displaying component widgets Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 49/59] jme: Do not enable NIC WoL functions on S0 Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 19/59] drm/i915: Fix system resume if PCI device remained enabled Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 54/59] batman-adv: Fix broadcast/ogm queue limit on a removed interface Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 59/59] RDMA/iw_cxgb4: Fix bar2 virt addr calculation for T4 chips Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 32/59] EDAC: i7core, sb_edac: Don't return NOTIFY_BAD from mce_decoder callback Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 55/59] mm: update min_free_kbytes from khugepaged after core initialization Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 58/59] cxgbi: fix uninitialized flowi6 Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 56/59] ARM: EXYNOS: Properly skip unitialized parent clock in power domain on Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:00 +0200
[PATCH 4.2.y-ckt 24/59] drm/radeon: fix vertical bars appear on monitor (v2) Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 33/59] powerpc: Fix bad inline asm constraint in create_zero_mask() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 42/59] x86/sysfb_efi: Fix valid BAR address range check Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 48/59] parisc: fix a bug when syscall number of tracee is __NR_Linux_syscalls Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 41/59] ARC: Add missing io barriers to io{read,write}{16,32}be() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 45/59] writeback: Fix performance regression in wb_over_bg_thresh() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 35/59] drm/amdgpu: set metadata pointer to NULL after freeing. Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 34/59] Minimal fix-up of bad hashing behavior of hash_64() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 21/59] drm/i915: Fix eDP low vswing for Broadwell Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 44/59] propogate_mnt: Handle the first propogated copy being a slave Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 51/59] net/mlx4_en: fix spurious timestamping callbacks Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 30/59] mm/huge_memory: replace VM_NO_THP VM_BUG_ON with actual VMA check Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 26/59] x86/irq: Fix a race in x86_vector_free_irqs() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 43/59] fs/pnode.c: treat zero mnt_group_id-s as unequal Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 39/59] MAINTAINERS: Remove asterisk from EFI directory names Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 28/59] ARM: cpuidle: Pass on arm_cpuidle_suspend()'s return value Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 29/59] IB/security: Restrict use of the write() interface Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 47/59] x86/tsc: Read all ratio bits from MSR_PLATFORM_INFO Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 22/59] drm/i915: Make RPS EI/thresholds multiple of 25 on SNB-BDW Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 52/59] batman-adv: Reduce refcnt of removed router when updating route Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 46/59] mm, cma: prevent nr_isolated_* counters from going negative Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 31/59] mm: vmscan: reclaim highmem zone if buffer_heads is over limit Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 50/59] jme: Fix device PM wakeup API usage Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 36/59] tracing: Don't display trigger file for events that can't be enabled Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 38/59] drm/amdgpu: make sure vertical front porch is at least 1 Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 27/59] x86/apic: Handle zero vector gracefully in clear_vector_irq() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 40/59] ACPICA: Dispatcher: Update thread ID for recursive method calls Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:10 +0200
[PATCH 4.2.y-ckt 18/59] cxl: Keep IRQ mappings on context teardown Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 02/59] USB: usbip: fix potential out-of-bounds write Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 07/59] iio: ak8975: fix maybe-uninitialized warning Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 12/59] USB: serial: cp210x: add Straizona Focusers device ids Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 13/59] [media] v4l2-dv-timings.h: fix polarity for 4k formats Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 14/59] ALSA: hda - Add dock support for ThinkPad X260 Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 05/59] ath9k: ar5008_hw_cmn_spur_mitigate: add missing mask_m & mask_p initialisation Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 09/59] i2c: exynos5: Fix possible ABBA deadlock by keeping I2C clock prepared Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 20/59] drm/i915/ddi: Fix eDP VDD handling during booting and suspend/resume Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 06/59] iio: ak8975: Fix NULL pointer exception on early interrupt Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 01/59] x86/mm/32: Enable full randomization on i386 and X86_32 Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 16/59] drm/dp/mst: Get validated port ref in drm_dp_update_payload_part1() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 11/59] USB: serial: cp210x: add ID for Link ECU Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 10/59] efi: Fix out-of-bounds read in variable_matches() Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 15/59] workqueue: fix ghost PENDING flag while doing MQ IO Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
[PATCH 4.2.y-ckt 03/59] ASoC: rt5640: Correct the digital interface data select Kamal Mostafa <kamal@canonical.com> - 2016-05-09 22:20 +0200
csiph-web