♟ mhorne

libpmc: userland support and pmc.rapl.3 for the RAPL class

In D58027#1342622, @afscoelho_gmail.com wrote:

In D58027#1342399, @aokblast wrote:

Ok! I would like to confirm something with you before committing:

Author info: Andre Silva <andasilv@amd.com>

Sponsored by: Advanced Micro Devices (AMD)

Are the both desired information you expected?

Yes, they are. Thanks!

hwpmc: add RAPL energy-counter class (AMD + Intel)

pmc_dmc620: drop unneeded includes

x86: add AMD RAPL MSR definitions to specialreg.h

sys/modules/hwpmc/Makefile: tidy up

libsysdecode: fix nlm_flag regex in mktables

Thanks. I was considering removing all this code, but it was premature. I am happy to keep it if it is useful.

hwpmc: Add EXTERROR diagnostics to the AMD and IBS allocators

hwpmc: Add EXTERROR diagnostics to the hwpmc syscall path

vfs_unmountall.9: Fix an outdated .Xr to boot(9)

Seems sensible as a starting point.

Per the replies in https://github.com/freebsd/freebsd-src/pull/2245, this ends up having no functional impact. Closing.

So, this applies the EOI handling scheme preferred by many drivers, and endorsed by mmel@.

Revert "rk_gpio: defer level-IRQ EOI until source line is driven low"

subr_physmem_test: add tests for two edge-cases

In D45914#1335161, @markj wrote:

Rather late, but the patch seems fine to me.

g_eli: disambiguate CPU-bound worker creation

sched/lapic: remove sched_do_timer_accounting()

smp: remove unused hlt_cpus_mask

g_eli: better handling of absent/disabled CPUs

rk_gpio: defer level-IRQ EOI until source line is driven low

Obtain mutex around sc_workers list.

kern/sched_shims.c: back to ifunc

build: Stop testing LINKER_FEATURES for ifunc and build-id

arm: Implement kernel ifunc

DEFINE_IFUNC.9: update NOTES

sched_shim: Drop the no-ifunc case

riscv: implement kernel ifunc resolution

rk_gpio: implement PIC masking methods and mask unhandled IRQs

crashinfo: Create core.txt.last symlink

hwpmc_amd: Avoid using PMCs if in use by firmware

hwpmc: Disable AMD PMCs if in an unsupported VM

hwpmc_amd: fix amd_get_msr() MSR offset for newer counter bases

hwpmc_amd: Add support for additional counters.

hwpmc: Fix PMC flags for AMD Zen cores

LGTM. Ali, I will follow your lead that this is the most useful way to expose these counters.

LGTM. There was a recent addition to if_eqos_starfive to handle "local-mac-address", but it will be better to use this. Thanks.

sysutils/opensbi: update to v1.9

generic_ehci_fdt: fix driver softc size

stand/libofw: make OF_hasprop() part of the library

raspberry_virtgpio: fix OF_hasprop() usage

OF_getprop.9: update OF_hasprop() signature

clk_fixed: quiet a warning message

ofw: bool-ify OF_hasprop()

Hi, commenting here. From a high-level perspective this seems like a good move; this pmc(8) utility is open to improvements and seems like it never got past a beta phase.
It is good to be thoughtful about the old commands, but as you note they are broken and the tool as a whole is undocumented. So I do not think we are burdened with supporting them indefinitely.

generic_ehci_fdt: fix driver softc size

In D57381#1328509, @mhorne wrote:

I tested in QEMU (both Sv39 and Sv48 mode), and on the VisionFive v2 and Allwinner D1 hardware. In fact, it detected a real bug on the latter!

Hi. This is in great shape.

Is this implemented on the Spacemit K1 ?

hwpmc_ibs: Add more IBS register definitions

hwpmc_amd: Avoid using PMCs if in use by firmware

bcm2835_sdhci: Clean up DMA resources on attach failure

linux/io: handle memtype_wc mapping for !DMAP range

mhorne (Mitchell Horne)
User

Projects

User Details

Recent Activity
View All

Yesterday

Mon, Jul 27

Tue, Jul 21

Mon, Jul 20

Fri, Jul 17

Thu, Jul 16

Tue, Jul 14

Mon, Jul 13

Sun, Jul 12

Fri, Jul 10

Thu, Jul 9

Tue, Jun 30

Jun 29 2026

Jun 24 2026

Jun 23 2026

Jun 11 2026

mhorne (Mitchell Horne)User