Graphics driver (i915/Xe) on mordern platforms splits GFX and SA Media
information via different sysfs knobs.
Existing BIC_GFX_rc6/BIC_GFXMHz/BIC_GFXACTMHz columns can be reused for
GFX.
Introduce BIC_SAM_mc6/BIC_SAMMHz/BIC_SAMACTMHz columns for SA Media.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Running turbostat on a 16 socket HPE Scale-up Compute 3200 (SapphireRapids) fails with:
turbostat: /sys/devices/system/cpu/intel_uncore_frequency/package_010_die_00/current_freq_khz: open failed: No such file or directory
We observe the sysfs uncore frequency directories named:
...
package_09_die_00/
package_10_die_00/
package_11_die_00/
...
package_15_die_00/
The culprit is an incorrect sprintf format string "package_0%d_die_0%d" used
with each instance of reading uncore frequency files. uncore-frequency-common.c
creates the sysfs directory with the format "package_%02d_die_%02d". Once the
package value reaches double digits, the formats diverge.
Change each instance of "package_0%d_die_0%d" to "package_%02d_die_%02d".
[lenb: deleted the probe part of this patch, as it was already fixed]
Signed-off-by: Justin Ernst <justin.ernst@hpe.com>
Reviewed-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Len Brown <len.brown@intel.com>
Graphics sysfs snapshots share similar logic.
Combine them into one function to avoid code duplication.
No functional change.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Graphics drivers (i915/Xe) have different sysfs knobs on different
platforms, and it is possible that different sysfs knobs fit into the
same turbostat columns.
Instead of specifying different sysfs knobs every time, detect them
once and cache the path for future use.
No functional change.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some of the future Intel platforms will require reading the RAPL
counters via perf and not MSR. On current platforms we can still read
them using both ways.
Signed-off-by: Patryk Wlazlyn <patryk.wlazlyn@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
If user request --no-msr or is not able to access the MSRs,
turbostat should clear all the counters added with --add.
Because MSR access permission checks are done after the cmdline is
parsed, the decision has to be defered up until the transition into
no-msr mode happen.
Signed-off-by: Len Brown <len.brown@intel.com>
Checking early if the permissions are even needed gets rid of the
warnings about some of them missing. Earlier we issued a warning in case
of missing MSR and/or perf permissions, even when user never asked for
counters that require those.
Signed-off-by: Patryk Wlazlyn <patryk.wlazlyn@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
By using the perf API we spend less time in between the reads of the
counters, resulting in more accurate calculations of the dependent
metrics.
Using perf API is also usually faster overall, although cache miss, if
we get one, is more costly when using perf vs MSR driver.
We would fallback to the msr reads if the sysfs isn't there or when in
--no-perf mode.
Signed-off-by: Patryk Wlazlyn <patryk.wlazlyn@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Eliminate redundant debug output for core and package scope counters.
Include name and path for all "ADDED" counters.
Signed-off-by: Len Brown <len.brown@intel.com>
Previously a failed read of /dev/cpu_dma_latency erroneously complained
turbostat: capget(CAP_SYS_ADMIN) failed, try "# setcap cap_sys_admin=ep ./turbostat
This went unnoticed because this file is typically visible to root,
and turbostat was typically run as root.
Going forward, when a non-root user can run turbostat...
Complain about failed read access to this file only if --debug is used.
Signed-off-by: Len Brown <len.brown@intel.com>
Print current frequency along with the current (and initial) limits
Probe and print uncore config also for machines using the new cluster API
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat prints the abnormal SYS%LPI across suspend-to-idle:
SYS%LPI = 114479815993277.50
This is reproduced by:
Run a freeze cycle, e.g. "sleepgraph -m freeze -rtcwake 15".
Then do a reboot. After boot up, launch the suspend-idle-idle
and check the SYS%LPI field.
The slp_so residence counter is in LPIT table, and BIOS does not
clears this register across reset. The PMC expects the OS to calculate
the LPI residency based on the delta. However, there is an firmware
issue that the LPIT gets cleared to 0 during the second suspend
to idle after the reboot, which brings negative delta value.
[lenb: updated to print "neg" upon this BIOS failure]
Reported-by: Todd Brandt <todd.e.brandt@intel.com>
Signed-off-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The code calculates Bzy_MHz by multiplying TSC_delta * APERF_delta/MPERF_delta
The man page erroneously showed that TSC_delta was divided.
Signed-off-by: Peng Liu <liupeng17@lenovo.com>
Signed-off-by: Len Brown <len.brown@intel.com>
When running turbostat, a system with 512 cpus reaches the limit for
maximum number of file descriptors that can be opened. To solve this
problem, the limit is raised to 2^15, which is a large enough number.
Below data is collected from AMD server systems while running turbostat:
|-----------+-------------------------------|
| # of cpus | # of opened fds for turbostat |
|-----------+-------------------------------|
| 128 | 260 |
|-----------+-------------------------------|
| 192 | 388 |
|-----------+-------------------------------|
| 512 | 1028 |
|-----------+-------------------------------|
So, the new max limit would be sufficient up to 2^14 cpus (but this
also depends on how many counters are enabled).
Reviewed-by: Doug Smythies <dsmythies@telus.net>
Tested-by: Doug Smythies <dsmythies@telus.net>
Signed-off-by: Wyes Karny <wyes.karny@amd.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Pull i2c fixes from Wolfram Sang:
"Improve the detection when to run atomic transfer handlers for kernels
with preemption disabled. This removes some false positive splats a
number of users were seeing if their driver didn't have support for
atomic transfers.
Also, fix a typo in the docs while we are here"
* tag 'i2c-for-6.7-final' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
i2c: core: Fix atomic xfer check for non-preempt config
Documentation/i2c: fix spelling error in i2c-address-translators
Pull misc mm fixes from Andrew Morton:
"12 hotfixes.
Two are cc:stable and the remainder either address post-6.7 issues or
aren't considered necessary for earlier kernel versions"
* tag 'mm-hotfixes-stable-2024-01-05-11-35' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
mm: shrinker: use kvzalloc_node() from expand_one_shrinker_info()
mailmap: add entries for Mathieu Othacehe
MAINTAINERS: change vmware.com addresses to broadcom.com
arch/mm/fault: fix major fault accounting when retrying under per-VMA lock
mm/mglru: skip special VMAs in lru_gen_look_around()
MAINTAINERS: hand over hwpoison maintainership to Miaohe Lin
MAINTAINERS: remove hugetlb maintainer Mike Kravetz
mm: fix unmap_mapping_range high bits shift bug
mm: memcg: fix split queue list crash when large folio migration
mm: fix arithmetic for max_prop_frac when setting max_ratio
mm: fix arithmetic for bdi min_ratio
mm: align larger anonymous mappings on THP boundaries
Pull nfsd fix from Chuck Lever:
- Fix another regression in the NFSD administrative API
* tag 'nfsd-6.7-3' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
nfsd: drop the nfsd_put helper
Pull firewire fix from Takashi Sakamoto:
"A single patch to suppress unexpected system reboot in AMD Ryzen
machines with PCIe card consisting of Asmedia ASM1083/1085 and
VT6306/6307/6308.
When the 1394 OHCI driver for the card accesses a specific register
in PCI memory space, the system reboot often occurs.
The issue affects all versions of Linux kernel as long as the 1394
OHCI driver is included. The mechanism of unexpected system reboot is
not clear, so the driver is changed to avoid the access itself when
detecting the combination of hardware"
* tag 'firewire-fixes-6.7-final' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394:
firewire: ohci: suppress unexpected system reboot in AMD Ryzen machines and ASM108x/VT630x PCIe cards
Pull MMC fixes from Ulf Hansson:
"MMC core:
- Fix releasing the host by canceling the delayed work
- Fix pause retune on all RPMB partitions
MMC host:
- meson-mx-sdhc: Fix HW hang during card initialization
- sdhci-sprd: Fix eMMC init failure after HW reset"
* tag 'mmc-v6.7-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
mmc: sdhci-sprd: Fix eMMC init failure after hw reset
mmc: core: Cancel delayed work before releasing host
mmc: rpmb: fixes pause retune on all RPMB partitions.
mmc: meson-mx-sdhc: Fix initialization frozen issue
Pull more drm fixes from Dave Airlie:
"The amdgpu ones are fairly normal, the one that is a bit large is a
fix for a newly introduced IP in 6.7 so unlikely to cause regressions.
The nouveau ones are mostly memory leaks and debugging cleanups from
the GSP (new nvidia firmware) enablement. There are some GSP changes
to the message passing code and a subsequent fix for eDP panel turn
on, that means my laptop can turn on the panel in GSP mode. These are
fairly low chance of disrupting things since GSP is new in 6.7. The
final not all in GSP fix is a deadlock seen with i915/nouveau when GSP
is used where the the fence and irq paths have locking inversions,
I've pushed some irq enablement out to a workqueue, and this has seen
some fairly decent testing.
amdgpu:
- DP MST fix
- SMU 13.0.6 fixes
- fix displays on macbooks using vega12
- fix VSC and colorimetry on DP/eDP
nouveau:
- fix deadlock between fence signalling and irq paths
- fix GSP memory leaks
- fix GSP leftover debug
- hide some GSP callback messages
- fix GSP display disable path
- fix GSP ACPI interaction
- handle errors in ctrl messages
- use errors info to fix DP link training"
* tag 'drm-fixes-2024-01-05' of git://anongit.freedesktop.org/drm/drm:
drm/nouveau/dp: Honor GSP link training retry timeouts
nouveau: push event block/allowing out of the fence context
nouveau/gsp: always free the alloc messages on r535
nouveau/gsp: don't free ctrl messages on errors
nouveau/gsp: convert gsp errors to generic errors
drm/nouveau/gsp: Fix ACPI MXDM/MXDS method invocations
nouveau/gsp: free userd allocation.
nouveau/gsp: free acpi object after use
nouveau: fix disp disabling with GSP
nouveau/gsp: drop some acpi related debug
nouveau/gsp: add three notifier callbacks that we see in normal operation (v2)
drm/amd/pm: Use gpu_metrics_v1_5 for SMUv13.0.6
drm/amd/pm: Add gpu_metrics_v1_5
drm/amd/pm: Add mem_busy_percent for GCv9.4.3 apu
drm/amd/display: Fix sending VSC (+ colorimetry) packets for DP/eDP displays without PSR
drm/amdgpu: skip gpu_info fw loading on navi12
drm/amd/display: add nv12 bounding box
drm/amd/pm: Update metric table for jpeg/vcn data
drm/amd/pm: Use separate metric table for APU
drm/amd/display: pbn_div need be updated for hotplug event
Pull ARM SoC fixes from Arnd Bergmann:
"These are two correctness fixes for handing DT input in the
Allwinner (sunxi) SMP startup code"
* tag 'soc-fixes-6.7-3a' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
ARM: sun9i: smp: fix return code check of of_property_match_string
ARM: sun9i: smp: Fix array-index-out-of-bounds read in sunxi_mc_smp_init
Pull kvm fix from Paolo Bonzini:
- Fix boolean logic in intel_guest_get_msrs
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: x86/pmu: fix masking logic for MSR_CORE_PERF_GLOBAL_CTRL
Pull kprobes/x86 fix from Masami Hiramatsu:
- Fix to emulate indirect call which size is not 5 byte.
Current code expects the indirect call instructions are 5 bytes, but
that is incorrect. Usually indirect call based on register is shorter
than that, thus the emulation causes a kernel crash by accessing
wrong instruction boundary. This uses the instruction size to
calculate the return address correctly.
* tag 'probes-fixes-v6.7-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
x86/kprobes: fix incorrect return address calculation in kprobe_emulate_call_indirect
Pull smb client fixes from Steve French:
"Three important multichannel smb3 client fixes found in recent
testing:
- fix oops due to incorrect refcounting of interfaces after
disabling multichannel
- fix possible unrecoverable session state after disabling
multichannel with active sessions
- fix two places that were missing use of chan_lock"
* tag '6.7-rc8-smb3-mchan-fixes' of git://git.samba.org/sfrench/cifs-2.6:
cifs: do not depend on release_iface for maintaining iface_list
cifs: cifs_chan_is_iface_active should be called with chan_lock held
cifs: after disabling multichannel, mark tcon for reconnect
VIA VT6306/6307/6308 provides PCI interface compliant to 1394 OHCI. When
the hardware is combined with Asmedia ASM1083/1085 PCIe-to-PCI bus bridge,
it appears that accesses to its 'Isochronous Cycle Timer' register (offset
0xf0 on PCI memory space) often causes unexpected system reboot in any
type of AMD Ryzen machine (both 0x17 and 0x19 families). It does not
appears in the other type of machine (AMD pre-Ryzen machine, Intel
machine, at least), or in the other OHCI 1394 hardware (e.g. Texas
Instruments).
The issue explicitly appears at a commit dcadfd7f7c ("firewire: core:
use union for callback of transaction completion") added to v6.5 kernel.
It changed 1394 OHCI driver to access to the register every time to
dispatch local asynchronous transaction. However, the issue exists in
older version of kernel as long as it runs in AMD Ryzen machine, since
the access to the register is required to maintain bus time. It is not
hard to imagine that users experience the unexpected system reboot when
generating bus reset by plugging any devices in, or reading the register
by time-aware application programs; e.g. audio sample processing.
This commit suppresses the unexpected system reboot in the combination of
hardware. It avoids the access itself. As a result, the software stack can
not provide the hardware time anymore to unit drivers, userspace
applications, and nodes in the same IEEE 1394 bus. It brings apparent
disadvantage since time-aware application programs require it, while
time-unaware applications are available again; e.g. sbp2.
Cc: stable@vger.kernel.org
Reported-by: Jiri Slaby <jirislaby@kernel.org>
Closes: https://bugzilla.suse.com/show_bug.cgi?id=1215436
Reported-by: Mario Limonciello <mario.limonciello@amd.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217994
Reported-by: Tobias Gruetzmacher <tobias-lists@23.gs>
Closes: https://sourceforge.net/p/linux1394/mailman/message/58711901/
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=2240973
Closes: https://bugs.launchpad.net/linux/+bug/2043905
Link: https://lore.kernel.org/r/20240102110150.244475-1-o-takashi@sakamocchi.jp
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
It's not safe to call nfsd_put once nfsd_last_thread has been called, as
that function will zero out the nn->nfsd_serv pointer.
Drop the nfsd_put helper altogether and open-code the svc_put in its
callers instead. That allows us to not be reliant on the value of that
pointer when handling an error.
Fixes: 2a501f55cd ("nfsd: call nfsd_last_thread() before final nfsd_put()")
Reported-by: Zhi Li <yieli@redhat.com>
Cc: NeilBrown <neilb@suse.de>
Signed-off-by: Jeffrey Layton <jlayton@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Turns out that one of the ways that Nvidia's driver handles the pre-LT
timeout for eDP panels is by providing a retry timeout in their link
training callbacks that we're expected to wait for. Up until now we didn't
pay any attention to this parameter.
So, start honoring the timeout if link training fails - and retry up to 3
times. The "3 times" bit comes from OpenRM's link training code.
[airlied: this fixes the panel on one of my laptops]
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231222043308.3090089-12-airlied@gmail.com
There is a deadlock between the irq and fctx locks,
the irq handling takes irq then fctx lock
the fence signalling takes fctx then irq lock
This splits the fence signalling path so the code that hits
the irq lock is done in a separate work queue.
This seems to fix crashes/hangs when using nouveau gsp with
i915 primary GPU.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231222043308.3090089-11-airlied@gmail.com
Currently we get an error from ACPI because both of these arguments expect
a single argument, and we don't provide one. I'm not totally clear on what
that argument does, but we're able to find the missing value from
_acpiCacheMethodData() in src/kernel/platform/acpi_common.c in nvidia's
driver. So, let's add that - which doesn't get eDP displays to power on
quite yet, but gets rid of the argument warning at least.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231222043308.3090089-7-airlied@gmail.com
