| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The system will be hang up during S3 suspend because of SMU is pending
for GC not respose the register CP_HQD_ACTIVE access request.This issue
root cause of accessing the GC register under enter GFX CGGPG and can
be fixed by disable GFX CGPG before perform suspend.
v2: Use disable the GFX CGPG instead of RLC safe mode guard.
Signed-off-by: Prike Liang <[email protected]>
Tested-by: Mengbing Wang <[email protected]>
Reviewed-by: Huang Rui <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
[WHY]
In cases where a clock table is malformed such that fclk entries have
frequencies but not voltages listed, we don't catch the error and set
clocks to 0 instead of using hardcoded values as we should.
[HOW]
Add check for clock tables fclk entry's voltage as well
Signed-off-by: Michael Strauss <[email protected]>
Reviewed-by: Eric Yang <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
[Why]
If dc->clk_mgr->funcs->are_clock_states_equal is set, then
wm_optimized_required is never checked. In that case, when going from a
higher mode to a lower mode, wm_optimized_required remains true until
the next mode change.
[How]
- move from else-if to unconditional or
Signed-off-by: Joshua Aberback <[email protected]>
Reviewed-by: Jun Lei <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Yongqiang Sun <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
In some usecases, like tiled display, the stream and plane configuration
can be setup in a way where the caller expects DAL to perform the
clipping, eg:
P0:
src_rect(0, 0, w, h)
dst_rect(0, 0, w, h)
P1:
src_rect(w, 0, w, h)
dst_rect(0, 0, w, h)
Cursor is enabled on both streams with the same position.
This can result in double cursor on tiled display, even though this
behavior is technically correct from the DC interface point of view.
We need a mechanism to control this dynamically.
[How]
This is something that should live in the DM layer based on detection
of the specified configuration but it's not something that we really
have enough information to deal with today.
Add a flag to the cursor position state that specifies whether we
want DC to do the translation or not and make it opt-in and let
the DM decide when to do it.
Signed-off-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Tony Cheng <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
If a plane isn't being actively enabled or disabled then DC won't
always recalculate scaling rects and ratios for the primary plane.
This results in only a partial or corrupted rect being displayed on
the screen instead of scaling to fit the screen.
[How]
Add back the logic to recalculate the scaling rects into
dc_commit_updates_for_stream since this is the expected place to
do it in DC.
This was previously removed a few years ago to fix an underscan issue
but underscan is still functional now with this change - and it should
be, since this is only updating to the latest plane state getting passed
in.
Signed-off-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Aric Cyr <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
For medium updates that change nothing but the source rect position
the viewport doesn't change on DCN20.
We're missing the check for the position update bit that was there in
the DCN10 hardware sequencer.
[How]
Check the position bit along with the scaling bit like we were doing
with DCN20.
We shouldn't actually hit a case where context != current_state in
our programming/commit model but guard against it anyway since it was
guarded for the other bits.
Signed-off-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Zhan Liu <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Cursor pos is correctly adjusted from DC side for source rect offset
on DCN ASIC, but only on the overlay.
This is because DM places offsets the cursor for primary planes only
to workaround missing code in DCE for the adjustment we're now correctly
doing in DC for DCN ASIC.
[How]
Drop the adjustment for source rect from the DM side of things and put
the code where it actually belongs - in DC on the pipe level.
This matches what we do for DCN now.
Signed-off-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Zhan Liu <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Changing policy to dynamic will allow 4k multi display configs
to be supported at DPM0
Signed-off-by: Eric Yang <[email protected]>
Reviewed-by: Tony Cheng <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Cursor is drawn as part of the framebuffer for a plane on AMD hardware.
The cursor position on the framebuffer does not change even if the
source rect viewport for the cursor does. This causes the cursor to be
clipped.
The following IGT tests fail as a result of this issue:
- kms_plane_cursor@pipe-*-viewport-size-*
[How]
Offset cursor position by plane source rect viewport. If the viewport
is unscaled then the cursor is now correctly positioned on any
plane - primary or overlay.
There is still a hardware limitation for dealing with the cursor size
being incorrectly scaled but that's not something we can address.
Add some documentation explaining some of this in the code while we're
at it.
Signed-off-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Zhan Liu <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
After v_total_min and max are updated in vrr structure, the changes are
not reflected in stream adjust. When these values are read from stream
adjust it does not reflect the actual state of the system.
[How]
Set stream adjust values equal to vrr adjust values after vrr adjust
values are updated.
Signed-off-by: Isabel Zhang <[email protected]>
Reviewed-by: Alvin Lee <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Prop are created at boot stage, and not allowed to create new prop
after device registration.
[How]
Reuse the connector property from SST if exist.
Signed-off-by: Jerry (Fangzhi) Zuo <[email protected]>
Reviewed-by: Hersen Wu <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Replace dev_warn() with dev_info() and note that they are
optional to avoid confusing users.
The RAS TAs only exist on server boards and the HDCP and DTM
TAs only exist on client boards. They are optional either way.
Acked-by: Nirmoy Das <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
prefix RAS error related dmesg print with pci device info
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: John Clements <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
upon receiving uncorrectable error, query every GPU node for ras errors
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: John Clements <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Incorrect CG sequence will cause gfx timedout,
if we keep switching power profile mode
(enter profile mod such as PEAK will disable CG,
exit profile mode EXIT will enable CG)
when run Vulkan test case(case used for test: vkexample).
Signed-off-by: Chengming Gui <[email protected]>
Reviewed-by: Kenneth Feng <[email protected]>
Acked-by: Evan Quan <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
On ARCTURUS and RENOIR, powerplay is not supported yet.
When plug in or unplug power jack, ACPI event will issue.
Then kernel NULL pointer BUG will be triggered.
Check for NULL pointers before calling.
Signed-off-by: Aaron Ma <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
Change SH_MEM_CONFIG Alignment mode to Automatic, as:
1)OGL fn_amd_compute_shader will failed with unaligned mode.
2)The default alignment mode was defined to automatic on gfx10
specification.
Signed-off-by: Likun Gao <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
hwmgr->pm_en is initialized at hwmgr_hw_init.
during amdgpu_device_init, there is amdgpu_asic_reset that calls to
pp_get_asic_baco_capability, while hwmgr->pm_en has not yet been initialized.
so avoid using pm_en in pp_get_asic_baco_capability.
Reviewed-by: Emily Deng <[email protected]>
Signed-off-by: Tiecheng Zhou <[email protected]>
Signed-off-by: Yintian Tao <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
As the pmfw hasn't exported the interface of SMU feature
mask to APU SKU so just force on all the features to driver
inquired interface at early initial stage.
Signed-off-by: Prike Liang <[email protected]>
Reviewed-by: Huang Rui <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
Fixes build error of:
"use of undeclared identifier 'RENOIR_A0'"
To fix the same, this patch re-orders the
ASIC declarations accordingly.
Fixes: 41ef3dcd86443fa ("drm/amd/display: Fix RV2 Variant Detection")
Signed-off-by: Shirish S <[email protected]>
Reviewed-by: Zhan Liu <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
1.Using the FCLK DPM table to set the MCLK for DPM states consist of
three entities:
FCLK
UCLK
MEMCLK
All these three clk change together, MEMCLK from FCLK, so use the fclk
frequency.
2.we should show the current working clock freqency from clock table metric
Signed-off-by: Yuxian Dai <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Reviewed-by: Huang Rui <[email protected]>
Reviewed-by: Kevin Wang <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
Originally, it kfrees the wrong pointer for mem_obj.
It would cause memory leak under stress test.
Signed-off-by: Jack Zhang <[email protected]>
Acked-by: Nirmoy Das <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Some displays have an issue where the hdcp chips are initialized after the
display has already lit up. This means we can sometimes authentication too early
and cause authentication failures.
This happens when HDCP is enabled and the display is power cycled. Normally we
will authenticate 2 seconds after the display is lit, but some displays need a
bit more time.
[How]
Increase delay to 3 second before we start authentication.
Signed-off-by: Bhawanpreet Lakha <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
-We need to cancel future callbacks/watchdogs events when a callback/watchdog event happens
[How]
-fix typo in event_callback()
-cancel callback, not watchdog
-cancel watchdog events in event_watchdog_timer().
Signed-off-by: Bhawanpreet Lakha <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
When content type property is set to 1. We should enable hdcp2.2 and if we cant
then stop. Currently the way it works in DC is that if we fail hdcp2, we will
try hdcp1 after.
[How]
Use link config to force disable hdcp1.4 when type1 is set.
Signed-off-by: Bhawanpreet Lakha <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This is ASIC specific and should be placed in _ppt.c of each ASIC.
Signed-off-by: Evan Quan <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This is already done in soc15.c. And this is really ASIC specific
and should not be placed here.
Signed-off-by: Evan Quan <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
DML expects num_states to exclude the duplicate state.
[How]
Set num_states to correct value to prevent array off-by-one error. Also
refactor max clock level code for diags.
Signed-off-by: Dmytro Laktyushkin <[email protected]>
Signed-off-by: George Shen <[email protected]>
Reviewed-by: Dmytro Laktyushkin <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
BT2020 is not supported in COLOR_ENCODING property of planes. Only
BT601 and BT709 was available.
[How]
Allow BT2020 as legit value in setting COLOR_ENCODING property.
Signed-off-by: Stylon Wang <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Nominal pixel clock and EDID information differ in precision so although
monitor reports maximum refresh is 2x minimum, LFC was not being
enabled.
[How]
Use minimum refresh rate as nominal/2 when EDID dictates that min
refresh = max refresh/2.
v2: squash in 64 bit divide fix
Signed-off-by: Aric Cyr <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
CTM was only supported at CRTC level and we need color space conversion
in linear space at plane level.
[How]
- Add plane-level CTM to dc interface
- Program plane-level CTM in DCN
Signed-off-by: Stylon Wang <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Change is causing a regression where the OPC app no longer functions
properly.
[How]
Revert the changelist causing the issue.
Signed-off-by: Isabel Zhang <[email protected]>
Reviewed-by: Yongqiang Sun <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
In some scenario like 1366x768 VSR enabled connected with a 4K monitor
and playing 4K video in clone mode, underflow will be observed due to
decrease dppclk when previouse surface scan isn't finished
[How]
In this use case, surface flip is switching between 4K and 1366x768,
1366x768 needs smaller dppclk, and when decrease the clk and previous
surface scan is for 4K and scan isn't done, underflow will happen. Not
doing optimize bandwidth in case of flip pending.
Signed-off-by: Yongqiang Sun <[email protected]>
Reviewed-by: Tony Cheng <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
For some monitors extreme flickering can occur while using LFC for if
we're not doing the DRR timing update for V_TOTAL_MIN / V_TOTAL_MAX at
the DP start of frame.
Hardware can default to any time in the frame which isn't the behavior
we want.
[How]
Add a new function for setting the double buffering mode for DRR timing.
Default to DP start of frame double buffering on timing generator init.
Signed-off-by: Nicholas Kazlauskas <[email protected]>
Reviewed-by: Aric Cyr <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
P010 pixel format is not declared as supported in DRM and DM.
[How]
Add P010 format to the support list presented to DRM and checked in DM
Signed-off-by: Stylon Wang <[email protected]>
Reviewed-by: Nicholas Kazlauskas <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
[Why]
Diagnostics scaling test failing to set required number of vertical taps
in 4:2:0 surface case
[How]
In dpp3_get_optimal_number_of_taps() need to use LB_MEMORY_CONFIG_3 for
4:2:0 surface case. In resource_build_scaling_params() make sure to also
set plane res alpha enable based on updated surface state
Signed-off-by: Eric Bernstein <[email protected]>
Reviewed-by: Dmytro Laktyushkin <[email protected]>
Acked-by: Rodrigo Siqueira <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
the HPD bo size calculation error.
the "mem.size" can't present actual BO size all time.
Signed-off-by: Kevin Wang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Acked-by: Christian König <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
https://gitlab.freedesktop.org/drm/msm into drm-next
A bit smaller this time around.. there are still a couple uabi
additions for vulkan waiting in the wings, but I punted on them this
cycle due to running low on time. (They should be easy enough to
rebase, and if it is a problem for anyone I can push a next+uabi
branch so that tu work can proceed.)
The bigger change is refactoring dpu resource manager and moving dpu
to use atomic global state. Other than that, it is mostly cleanups
and fixes.
From: Rob Clark <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/ <CAF6AEGuf1R4Xz-t9Z7_cwx9jD=b4wUvvwfqA5cHR8fCSXSd5XQ@mail.gmail.com
Signed-off-by: Dave Airlie <[email protected]>
|
|
msm needed rc6, so I just went and merged release
(msm has been in drm-next outside of this tree)
Signed-off-by: Dave Airlie <[email protected]>
|
|
git://anongit.freedesktop.org/drm/drm-intel into drm-next
Fixes for instability on Baytrail and Haswell;
Ice Lake RPS; Sandy Bridge RC6; and few others around
GT hangchec/reset; livelock; and a null dereference.
Signed-off-by: Dave Airlie <[email protected]>
From: Rodrigo Vivi <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
|
|
git://people.freedesktop.org/~agd5f/linux into drm-next
amd-drm-next-5.7-2020-03-26:
amdgpu:
- Remove a dpm quirk that is not necessary
- Fix handling of AC/DC mode in newer SMU firmwares on navi
- SR-IOV fixes
- RAS fixes
scheduler:
- Fix a race condition
radeon:
- Remove a dpm quirk that is not necessary
Signed-off-by: Dave Airlie <[email protected]>
From: Alex Deucher <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
|
|
|
|
Merge vm fixes from Andrew Morton:
"5 fixes"
* emailed patches from Andrew Morton <[email protected]>:
mm/sparse: fix kernel crash with pfn_section_valid check
mm: fork: fix kernel_stack memcg stats for various stack implementations
hugetlb_cgroup: fix illegal access to memory
drivers/base/memory.c: indicate all memory blocks as removable
mm/swapfile.c: move inode_lock out of claim_swapfile
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer fix from Thomas Gleixner:
"A single fix for the Hyper-V clocksource driver to make sched clock
actually return nanoseconds and not the virtual clock value which
increments at 10e7 HZ (100ns)"
* tag 'timers-urgent-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
clocksource/drivers/hyper-v: Make sched clock return nanoseconds correctly
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq fix from Thomas Gleixner:
"A single bugfix to prevent reference leaks in irq affinity notifiers"
* tag 'irq-urgent-2020-03-29' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
genirq: Fix reference leaks on irq affinity notifiers
|
|
Fix the crash like this:
BUG: Kernel NULL pointer dereference on read at 0x00000000
Faulting instruction address: 0xc000000000c3447c
Oops: Kernel access of bad area, sig: 11 [#1]
LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
CPU: 11 PID: 7519 Comm: lt-ndctl Not tainted 5.6.0-rc7-autotest #1
...
NIP [c000000000c3447c] vmemmap_populated+0x98/0xc0
LR [c000000000088354] vmemmap_free+0x144/0x320
Call Trace:
section_deactivate+0x220/0x240
__remove_pages+0x118/0x170
arch_remove_memory+0x3c/0x150
memunmap_pages+0x1cc/0x2f0
devm_action_release+0x30/0x50
release_nodes+0x2f8/0x3e0
device_release_driver_internal+0x168/0x270
unbind_store+0x130/0x170
drv_attr_store+0x44/0x60
sysfs_kf_write+0x68/0x80
kernfs_fop_write+0x100/0x290
__vfs_write+0x3c/0x70
vfs_write+0xcc/0x240
ksys_write+0x7c/0x140
system_call+0x5c/0x68
The crash is due to NULL dereference at
test_bit(idx, ms->usage->subsection_map);
due to ms->usage = NULL in pfn_section_valid()
With commit d41e2f3bd546 ("mm/hotplug: fix hot remove failure in
SPARSEMEM|!VMEMMAP case") section_mem_map is set to NULL after
depopulate_section_mem(). This was done so that pfn_page() can work
correctly with kernel config that disables SPARSEMEM_VMEMMAP. With that
config pfn_to_page does
__section_mem_map_addr(__sec) + __pfn;
where
static inline struct page *__section_mem_map_addr(struct mem_section *section)
{
unsigned long map = section->section_mem_map;
map &= SECTION_MAP_MASK;
return (struct page *)map;
}
Now with SPASEMEM_VMEMAP enabled, mem_section->usage->subsection_map is
used to check the pfn validity (pfn_valid()). Since section_deactivate
release mem_section->usage if a section is fully deactivated,
pfn_valid() check after a subsection_deactivate cause a kernel crash.
static inline int pfn_valid(unsigned long pfn)
{
...
return early_section(ms) || pfn_section_valid(ms, pfn);
}
where
static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
{
int idx = subsection_map_index(pfn);
return test_bit(idx, ms->usage->subsection_map);
}
Avoid this by clearing SECTION_HAS_MEM_MAP when mem_section->usage is
freed. For architectures like ppc64 where large pages are used for
vmmemap mapping (16MB), a specific vmemmap mapping can cover multiple
sections. Hence before a vmemmap mapping page can be freed, the kernel
needs to make sure there are no valid sections within that mapping.
Clearing the section valid bit before depopulate_section_memap enables
this.
[[email protected]: add comment]
Link: http://lkml.kernel.org/r/[email protected]: http://lkml.kernel.org/r/[email protected]
Fixes: d41e2f3bd546 ("mm/hotplug: fix hot remove failure in SPARSEMEM|!VMEMMAP case")
Reported-by: Sachin Sant <[email protected]>
Signed-off-by: Aneesh Kumar K.V <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Tested-by: Sachin Sant <[email protected]>
Reviewed-by: Baoquan He <[email protected]>
Reviewed-by: Wei Yang <[email protected]>
Acked-by: Michal Hocko <[email protected]>
Acked-by: Pankaj Gupta <[email protected]>
Cc: Michael Ellerman <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: David Hildenbrand <[email protected]>
Cc: Oscar Salvador <[email protected]>
Cc: Mike Rapoport <[email protected]>
Cc: <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
|
|
Depending on CONFIG_VMAP_STACK and the THREAD_SIZE / PAGE_SIZE ratio the
space for task stacks can be allocated using __vmalloc_node_range(),
alloc_pages_node() and kmem_cache_alloc_node().
In the first and the second cases page->mem_cgroup pointer is set, but
in the third it's not: memcg membership of a slab page should be
determined using the memcg_from_slab_page() function, which looks at
page->slab_cache->memcg_params.memcg . In this case, using
mod_memcg_page_state() (as in account_kernel_stack()) is incorrect:
page->mem_cgroup pointer is NULL even for pages charged to a non-root
memory cgroup.
It can lead to kernel_stack per-memcg counters permanently showing 0 on
some architectures (depending on the configuration).
In order to fix it, let's introduce a mod_memcg_obj_state() helper,
which takes a pointer to a kernel object as a first argument, uses
mem_cgroup_from_obj() to get a RCU-protected memcg pointer and calls
mod_memcg_state(). It allows to handle all possible configurations
(CONFIG_VMAP_STACK and various THREAD_SIZE/PAGE_SIZE values) without
spilling any memcg/kmem specifics into fork.c .
Note: This is a special version of the patch created for stable
backports. It contains code from the following two patches:
- mm: memcg/slab: introduce mem_cgroup_from_obj()
- mm: fork: fix kernel_stack memcg stats for various stack implementations
[[email protected]: introduce mem_cgroup_from_obj()]
Link: http://lkml.kernel.org/r/[email protected]
Fixes: 4d96ba353075 ("mm: memcg/slab: stop setting page->mem_cgroup pointer for slab pages")
Signed-off-by: Roman Gushchin <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Reviewed-by: Shakeel Butt <[email protected]>
Acked-by: Johannes Weiner <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Bharata B Rao <[email protected]>
Cc: Shakeel Butt <[email protected]>
Cc: <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Linus Torvalds <[email protected]>
|
|
This appears to be a mistake in commit faced7e0806cf ("mm: hugetlb
controller for cgroups v2").
Essentially that commit does a hugetlb_cgroup_from_counter assuming that
page_counter_try_charge has initialized counter.
But if that has failed then it seems will not initialize counter, so
hugetlb_cgroup_from_counter(counter) ends up pointing to random memory,
causing kasan to complain.
The solution is to simply use 'h_cg', instead of
hugetlb_cgroup_from_counter(counter), since that is a reference to the
hugetlb_cgroup anyway. After this change kasan ceases to complain.
Fixes: faced7e0806cf ("mm: hugetlb controller for cgroups v2")
Reported-by: [email protected]
Signed-off-by: Mina Almasry <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Acked-by: Giuseppe Scrivano <[email protected]>
Acked-by: Tejun Heo <[email protected]>
Cc: Mike Kravetz <[email protected]>
Cc: David Rientjes <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Linus Torvalds <[email protected]>
|
|
We see multiple issues with the implementation/interface to compute
whether a memory block can be offlined (exposed via
/sys/devices/system/memory/memoryX/removable) and would like to simplify
it (remove the implementation).
1. It runs basically lockless. While this might be good for performance,
we see possible races with memory offlining that will require at
least some sort of locking to fix.
2. Nowadays, more false positives are possible. No arch-specific checks
are performed that validate if memory offlining will not be denied
right away (and such check will require locking). For example, arm64
won't allow to offline any memory block that was added during boot -
which will imply a very high error rate. Other archs have other
constraints.
3. The interface is inherently racy. E.g., if a memory block is detected
to be removable (and was not a false positive at that time), there is
still no guarantee that offlining will actually succeed. So any
caller already has to deal with false positives.
4. It is unclear which performance benefit this interface actually
provides. The introducing commit 5c755e9fd813 ("memory-hotplug: add
sysfs removable attribute for hotplug memory remove") mentioned
"A user-level agent must be able to identify which sections
of memory are likely to be removable before attempting the
potentially expensive operation."
However, no actual performance comparison was included.
Known users:
- lsmem: Will group memory blocks based on the "removable" property. [1]
- chmem: Indirect user. It has a RANGE mode where one can specify
removable ranges identified via lsmem to be offlined. However,
it also has a "SIZE" mode, which allows a sysadmin to skip the
manual "identify removable blocks" step. [2]
- powerpc-utils: Uses the "removable" attribute to skip some memory
blocks right away when trying to find some to offline+remove.
However, with ballooning enabled, it already skips this
information completely (because it once resulted in many false
negatives). Therefore, the implementation can deal with false
positives properly already. [3]
According to Nathan Fontenot, DLPAR on powerpc is nowadays no longer
driven from userspace via the drmgr command (powerpc-utils). Nowadays
it's managed in the kernel - including onlining/offlining of memory
blocks - triggered by drmgr writing to /sys/kernel/dlpar. So the
affected legacy userspace handling is only active on old kernels. Only
very old versions of drmgr on a new kernel (unlikely) might execute
slower - totally acceptable.
With CONFIG_MEMORY_HOTREMOVE, always indicating "removable" should not
break any user space tool. We implement a very bad heuristic now.
Without CONFIG_MEMORY_HOTREMOVE we cannot offline anything, so report
"not removable" as before.
Original discussion can be found in [4] ("[PATCH RFC v1] mm:
is_mem_section_removable() overhaul").
Other users of is_mem_section_removable() will be removed next, so that
we can remove is_mem_section_removable() completely.
[1] http://man7.org/linux/man-pages/man1/lsmem.1.html
[2] http://man7.org/linux/man-pages/man8/chmem.8.html
[3] https://github.com/ibm-power-utilities/powerpc-utils
[4] https://lkml.kernel.org/r/[email protected]
Also, this patch probably fixes a crash reported by Steve.
http://lkml.kernel.org/r/CAPcyv4jpdaNvJ67SkjyUJLBnBnXXQv686BiVW042g03FUmWLXw@mail.gmail.com
Reported-by: "Scargall, Steve" <[email protected]>
Suggested-by: Michal Hocko <[email protected]>
Signed-off-by: David Hildenbrand <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Reviewed-by: Nathan Fontenot <[email protected]>
Acked-by: Michal Hocko <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: "Rafael J. Wysocki" <[email protected]>
Cc: Badari Pulavarty <[email protected]>
Cc: Robert Jennings <[email protected]>
Cc: Heiko Carstens <[email protected]>
Cc: Karel Zak <[email protected]>
Cc: <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Linus Torvalds <[email protected]>
|
|
claim_swapfile() currently keeps the inode locked when it is successful,
or the file is already swapfile (with -EBUSY). And, on the other error
cases, it does not lock the inode.
This inconsistency of the lock state and return value is quite confusing
and actually causing a bad unlock balance as below in the "bad_swap"
section of __do_sys_swapon().
This commit fixes this issue by moving the inode_lock() and IS_SWAPFILE
check out of claim_swapfile(). The inode is unlocked in
"bad_swap_unlock_inode" section, so that the inode is ensured to be
unlocked at "bad_swap". Thus, error handling codes after the locking now
jumps to "bad_swap_unlock_inode" instead of "bad_swap".
=====================================
WARNING: bad unlock balance detected!
5.5.0-rc7+ #176 Not tainted
-------------------------------------
swapon/4294 is trying to release lock (&sb->s_type->i_mutex_key) at: __do_sys_swapon+0x94b/0x3550
but there are no more locks to release!
other info that might help us debug this:
no locks held by swapon/4294.
stack backtrace:
CPU: 5 PID: 4294 Comm: swapon Not tainted 5.5.0-rc7-BTRFS-ZNS+ #176
Hardware name: ASUS All Series/H87-PRO, BIOS 2102 07/29/2014
Call Trace:
dump_stack+0xa1/0xea
print_unlock_imbalance_bug.cold+0x114/0x123
lock_release+0x562/0xed0
up_write+0x2d/0x490
__do_sys_swapon+0x94b/0x3550
__x64_sys_swapon+0x54/0x80
do_syscall_64+0xa4/0x4b0
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7f15da0a0dc7
Fixes: 1638045c3677 ("mm: set S_SWAPFILE on blockdev swap devices")
Signed-off-by: Naohiro Aota <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Tested-by: Qais Youef <[email protected]>
Reviewed-by: Andrew Morton <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Linus Torvalds <[email protected]>
|