| From: |
| Tony Luck <tony.luck-AT-intel.com> |
| To: |
| Borislav Petkov <bp-AT-alien8.de> |
| Subject: |
| [Updates to v17 PATCH 00/03] x86,fs/resctrl telemetry monitoring |
| Date: |
| Thu, 08 Jan 2026 09:42:24 -0800 |
| Message-ID: |
| <20260108174227.11307-1-tony.luck@intel.com> |
| Cc: |
| Fenghua Yu <fenghuay-AT-nvidia.com>, Reinette Chatre <reinette.chatre-AT-intel.com>, Maciej Wieczor-Retman <maciej.wieczor-retman-AT-intel.com>, Peter Newman <peternewman-AT-google.com>, James Morse <james.morse-AT-arm.com>, Babu Moger <babu.moger-AT-amd.com>, Drew Fustini <dfustini-AT-baylibre.com>, Dave Martin <Dave.Martin-AT-arm.com>, Chen Yu <yu.c.chen-AT-intel.com>, x86-AT-kernel.org, linux-kernel-AT-vger.kernel.org, patches-AT-lists.linux.dev, Tony Luck <tony.luck-AT-intel.com> |
Updates to v17 series[0] to address Boris' suggestion[1] to use
DO_ONCE() instead of open-coding with atomic operations to only execute
some code once.
The substantive changes are in patch 13 where the open-coded "only once"
logic has been removed from resctrl_arch_pre_mount() and replaced using
DO_ONCE_SLEEPABLE(resctrl_arch_pre_mount); in rdt_get_tree().
This change requires some trivial changes to patches 16 and 29 so that
they apply cleanly on top of the changes to resctrl_arch_pre_mount().
I dropped Reinette's Reviewed-by tag from patch 13 because of the
changes. But she commented[2]:
"Placing DO_ONCE_SLEEPABLE() in rdt_get_tree() is fine by me"
Signed-off-by: Tony Luck <tony.luck@intel.com>
-Tony
[0] https://lore.kernel.org/all/20251217172121.12030-1-tony.l...
[1] https://lore.kernel.org/all/20260105191711.GBaVwON5nZn-uO...
[2] https://lore.kernel.org/all/36239cc0-0a25-40a9-86d1-57236...
--
2.52.0
