From 3ae8508663372b93c5556a887e96ed0ca5df0711 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Sun, 6 Apr 2025 12:23:22 -0400 Subject: [PATCH 01/16] tools/power turbostat: Document GNR UncMHz domain convention Document that on Intel Granite Rapids Systems, Uncore domains 0-2 are CPU domains, and uncore domains 3-4 are IO domains. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.8 | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/power/x86/turbostat/turbostat.8 b/tools/power/x86/turbostat/turbostat.8 index 144565151e1e..e86493880c16 100644 --- a/tools/power/x86/turbostat/turbostat.8 +++ b/tools/power/x86/turbostat/turbostat.8 @@ -205,6 +205,7 @@ The system configuration dump (if --quiet is not used) is followed by statistics \fBUncMHz\fP per-package uncore MHz, instantaneous sample. .PP \fBUMHz1.0\fP per-package uncore MHz for domain=1 and fabric_cluster=0, instantaneous sample. System summary is the average of all packages. +Intel Granite Rapids systems use domains 0-2 for CPUs, and 3-4 for IO, with cluster always 0. For the "--show" and "--hide" options, use "UncMHz" to operate on all UMHz*.* as a group. .SH TOO MUCH INFORMATION EXAMPLE By default, turbostat dumps all possible information -- a system configuration header, followed by columns for all counters. -- 2.51.0 From f8b136ef2605c1bf62020462d10e35228760aa19 Mon Sep 17 00:00:00 2001 From: Zhang Rui Date: Wed, 19 Mar 2025 08:53:07 +0800 Subject: [PATCH 02/16] tools/power turbostat: Restore GFX sysfs fflush() call Do fflush() to discard the buffered data, before each read of the graphics sysfs knobs. Fixes: ba99a4fc8c24 ("tools/power turbostat: Remove unnecessary fflush() call") Signed-off-by: Zhang Rui Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index 70e17d4ad9b6..c9a34c16c7a8 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -6039,6 +6039,7 @@ int snapshot_graphics(int idx) int retval; rewind(gfx_info[idx].fp); + fflush(gfx_info[idx].fp); switch (idx) { case GFX_rc6: -- 2.51.0 From 994633894f208a0151baaee1688ab3c431912553 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Sun, 6 Apr 2025 12:53:18 -0400 Subject: [PATCH 03/16] tools/power turbostat: re-factor sysfs code Probe cpuidle "sysfs" residency and counts separately, since soon we will make one disabled on, and the other disabled off. Clarify that some BIC (build-in-counters) are actually "groups". since we're about to re-name some of those groups. no functional change. Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 31 ++++++++++++++++++--------- 1 file changed, 21 insertions(+), 10 deletions(-) diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index c9a34c16c7a8..df0391bedcde 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -273,10 +273,10 @@ struct msr_counter bic[] = { #define BIC_NMI (1ULL << 61) #define BIC_CPU_c1e (1ULL << 62) -#define BIC_TOPOLOGY (BIC_Package | BIC_Node | BIC_CoreCnt | BIC_PkgCnt | BIC_Core | BIC_CPU | BIC_Die) -#define BIC_THERMAL_PWR (BIC_CoreTmp | BIC_PkgTmp | BIC_PkgWatt | BIC_CorWatt | BIC_GFXWatt | BIC_RAMWatt | BIC_PKG__ | BIC_RAM__ | BIC_SysWatt) -#define BIC_FREQUENCY (BIC_Avg_MHz | BIC_Busy | BIC_Bzy_MHz | BIC_TSC_MHz | BIC_GFXMHz | BIC_GFXACTMHz | BIC_SAMMHz | BIC_SAMACTMHz | BIC_UNCORE_MHZ) -#define BIC_IDLE (BIC_Busy | BIC_sysfs | BIC_CPU_c1 | BIC_CPU_c3 | BIC_CPU_c6 | BIC_CPU_c7 | BIC_GFX_rc6 | BIC_Pkgpc2 | BIC_Pkgpc3 | BIC_Pkgpc6 | BIC_Pkgpc7 | BIC_Pkgpc8 | BIC_Pkgpc9 | BIC_Pkgpc10 | BIC_CPU_LPI | BIC_SYS_LPI | BIC_Mod_c6 | BIC_Totl_c0 | BIC_Any_c0 | BIC_GFX_c0 | BIC_CPUGFX | BIC_SAM_mc6 | BIC_Diec6) +#define BIC_GROUP_TOPOLOGY (BIC_Package | BIC_Node | BIC_CoreCnt | BIC_PkgCnt | BIC_Core | BIC_CPU | BIC_Die) +#define BIC_GROUP_THERMAL_PWR (BIC_CoreTmp | BIC_PkgTmp | BIC_PkgWatt | BIC_CorWatt | BIC_GFXWatt | BIC_RAMWatt | BIC_PKG__ | BIC_RAM__ | BIC_SysWatt) +#define BIC_GROUP_FREQUENCY (BIC_Avg_MHz | BIC_Busy | BIC_Bzy_MHz | BIC_TSC_MHz | BIC_GFXMHz | BIC_GFXACTMHz | BIC_SAMMHz | BIC_SAMACTMHz | BIC_UNCORE_MHZ) +#define BIC_GROUP_IDLE (BIC_Busy | BIC_sysfs | BIC_CPU_c1 | BIC_CPU_c3 | BIC_CPU_c6 | BIC_CPU_c7 | BIC_GFX_rc6 | BIC_Pkgpc2 | BIC_Pkgpc3 | BIC_Pkgpc6 | BIC_Pkgpc7 | BIC_Pkgpc8 | BIC_Pkgpc9 | BIC_Pkgpc10 | BIC_CPU_LPI | BIC_SYS_LPI | BIC_Mod_c6 | BIC_Totl_c0 | BIC_Any_c0 | BIC_GFX_c0 | BIC_CPUGFX | BIC_SAM_mc6 | BIC_Diec6) #define BIC_OTHER (BIC_IRQ | BIC_NMI | BIC_SMI | BIC_ThreadC | BIC_CoreTmp | BIC_IPC) #define BIC_DISABLED_BY_DEFAULT (BIC_USEC | BIC_TOD | BIC_APIC | BIC_X2APIC) @@ -2354,16 +2354,16 @@ unsigned long long bic_lookup(char *name_list, enum show_hide_mode mode) retval |= ~0; break; } else if (!strcmp(name_list, "topology")) { - retval |= BIC_TOPOLOGY; + retval |= BIC_GROUP_TOPOLOGY; break; } else if (!strcmp(name_list, "power")) { - retval |= BIC_THERMAL_PWR; + retval |= BIC_GROUP_THERMAL_PWR; break; } else if (!strcmp(name_list, "idle")) { - retval |= BIC_IDLE; + retval |= BIC_GROUP_IDLE; break; } else if (!strcmp(name_list, "frequency")) { - retval |= BIC_FREQUENCY; + retval |= BIC_GROUP_FREQUENCY; break; } else if (!strcmp(name_list, "other")) { retval |= BIC_OTHER; @@ -10260,7 +10260,7 @@ int is_deferred_skip(char *name) return 0; } -void probe_sysfs(void) +void probe_cpuidle_residency(void) { char path[64]; char name_buf[16]; @@ -10304,6 +10304,16 @@ void probe_sysfs(void) if (state < min_state) min_state = state; } +} + +void probe_cpuidle_counts(void) +{ + char path[64]; + char name_buf[16]; + FILE *input; + int state; + int min_state = 1024, max_state = 0; + char *sp; for (state = 10; state >= 0; --state) { @@ -10602,7 +10612,8 @@ skip_cgroup_setting: print_bootcmd(); } - probe_sysfs(); + probe_cpuidle_residency(); + probe_cpuidle_counts(); if (!getuid()) set_rlimit(); -- 2.51.0 From 6f110a5e4f9977c31ce76fefbfef6fd4eab6bfb7 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 6 Apr 2025 10:00:04 -0700 Subject: [PATCH 04/16] Disable SLUB_TINY for build testing ... and don't error out so hard on missing module descriptions. Before commit 6c6c1fc09de3 ("modpost: require a MODULE_DESCRIPTION()") we used to warn about missing module descriptions, but only when building with extra warnigns (ie 'W=1'). After that commit the warning became an unconditional hard error. And it turns out not all modules have been converted despite the claims to the contrary. As reported by Damian Tometzki, the slub KUnit test didn't have a module description, and apparently nobody ever really noticed. The reason nobody noticed seems to be that the slub KUnit tests get disabled by SLUB_TINY, which also ends up disabling a lot of other code, both in tests and in slub itself. And so anybody doing full build tests didn't actually see this failre. So let's disable SLUB_TINY for build-only tests, since it clearly ends up limiting build coverage. Also turn the missing module descriptions error back into a warning, but let's keep it around for non-'W=1' builds. Reported-by: Damian Tometzki Link: https://lore.kernel.org/all/01070196099fd059-e8463438-7b1b-4ec8-816d-173874be9966-000000@eu-central-1.amazonses.com/ Cc: Masahiro Yamada Cc: Jeff Johnson Fixes: 6c6c1fc09de3 ("modpost: require a MODULE_DESCRIPTION()") Signed-off-by: Linus Torvalds --- mm/Kconfig | 2 +- scripts/mod/modpost.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/mm/Kconfig b/mm/Kconfig index d3fb3762887b..e113f713b493 100644 --- a/mm/Kconfig +++ b/mm/Kconfig @@ -201,7 +201,7 @@ config KVFREE_RCU_BATCHED config SLUB_TINY bool "Configure for minimal memory footprint" - depends on EXPERT + depends on EXPERT && !COMPILE_TEST select SLAB_MERGE_DEFAULT help Configures the slab allocator in a way to achieve minimal memory diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c index 92627e8d0e16..be89921d60b6 100644 --- a/scripts/mod/modpost.c +++ b/scripts/mod/modpost.c @@ -1603,7 +1603,7 @@ static void read_symbols(const char *modname) } if (!get_modinfo(&info, "description")) - error("missing MODULE_DESCRIPTION() in %s\n", modname); + warn("missing MODULE_DESCRIPTION() in %s\n", modname); } for (sym = info.symtab_start; sym < info.symtab_stop; sym++) { -- 2.51.0 From ec4acd3166d8a7a03b059d01b9c6f11a658e833f Mon Sep 17 00:00:00 2001 From: Len Brown Date: Sun, 6 Apr 2025 14:29:57 -0400 Subject: [PATCH 05/16] tools/power turbostat: disable "cpuidle" invocation counters, by default Create "pct_idle" counter group, the sofware notion of residency so it can now be singled out, independent of other counter groups. Create "cpuidle" group, the cpuidle invocation counts. Disable "cpuidle", by default. Create "swidle" = "cpuidle" + "pct_idle". Undocument "sysfs", the old name for "swidle", but keep it working for backwards compatibilty. Create "hwidle", all the HW idle counters Modify "idle", enabled by default "idle" = "hwidle" + "pct_idle" (and now excludes "cpuidle") Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.8 | 12 +++++----- tools/power/x86/turbostat/turbostat.c | 34 +++++++++++++++++++++------ 2 files changed, 33 insertions(+), 13 deletions(-) diff --git a/tools/power/x86/turbostat/turbostat.8 b/tools/power/x86/turbostat/turbostat.8 index e86493880c16..b74ed916057e 100644 --- a/tools/power/x86/turbostat/turbostat.8 +++ b/tools/power/x86/turbostat/turbostat.8 @@ -100,7 +100,7 @@ The column name "all" can be used to enable all disabled-by-default built-in cou .PP \fB--show column\fP show only the specified built-in columns. May be invoked multiple times, or with a comma-separated list of column names. .PP -\fB--show CATEGORY --hide CATEGORY\fP Show and hide also accept a single CATEGORY of columns: "all", "topology", "idle", "frequency", "power", "sysfs", "other". +\fB--show CATEGORY --hide CATEGORY\fP Show and hide also accept a single CATEGORY of columns: "all", "topology", "idle", "frequency", "power", "cpuidle", "hwidle", "swidle", "other". "idle" (enabled by default), includes "hwidle" and "idle_pct". "cpuidle" (default disabled) includes cpuidle software invocation counters. "swidle" includes "cpuidle" plus "idle_pct". "hwidle" includes only hardware based idle residency counters. Older versions of turbostat used the term "sysfs" for what is now "swidle". .PP \fB--Dump\fP displays the raw counter values. .PP @@ -158,15 +158,15 @@ The system configuration dump (if --quiet is not used) is followed by statistics .PP \fBSMI\fP The number of System Management Interrupts serviced CPU during the measurement interval. While this counter is actually per-CPU, SMI are triggered on all processors, so the number should be the same for all CPUs. .PP -\fBC1, C2, C3...\fP The number times Linux requested the C1, C2, C3 idle state during the measurement interval. The system summary line shows the sum for all CPUs. These are C-state names as exported in /sys/devices/system/cpu/cpu*/cpuidle/state*/name. While their names are generic, their attributes are processor specific. They the system description section of output shows what MWAIT sub-states they are mapped to on each system. +\fBC1, C2, C3...\fP The number times Linux requested the C1, C2, C3 idle state during the measurement interval. The system summary line shows the sum for all CPUs. These are C-state names as exported in /sys/devices/system/cpu/cpu*/cpuidle/state*/name. While their names are generic, their attributes are processor specific. They the system description section of output shows what MWAIT sub-states they are mapped to on each system. These counters are in the "cpuidle" group, which is disabled, by default. .PP -\fBC1+, C2+, C3+...\fP The idle governor idle state misprediction statistics. Inidcates the number times Linux requested the C1, C2, C3 idle state during the measurement interval, but should have requested a deeper idle state (if it exists and enabled). These statistics come from the /sys/devices/system/cpu/cpu*/cpuidle/state*/below file. +\fBC1+, C2+, C3+...\fP The idle governor idle state misprediction statistics. Inidcates the number times Linux requested the C1, C2, C3 idle state during the measurement interval, but should have requested a deeper idle state (if it exists and enabled). These statistics come from the /sys/devices/system/cpu/cpu*/cpuidle/state*/below file. These counters are in the "cpuidle" group, which is disabled, by default. .PP -\fBC1-, C2-, C3-...\fP The idle governor idle state misprediction statistics. Inidcates the number times Linux requested the C1, C2, C3 idle state during the measurement interval, but should have requested a shallower idle state (if it exists and enabled). These statistics come from the /sys/devices/system/cpu/cpu*/cpuidle/state*/above file. +\fBC1-, C2-, C3-...\fP The idle governor idle state misprediction statistics. Inidcates the number times Linux requested the C1, C2, C3 idle state during the measurement interval, but should have requested a shallower idle state (if it exists and enabled). These statistics come from the /sys/devices/system/cpu/cpu*/cpuidle/state*/above file. These counters are in the "cpuidle" group, which is disabled, by default. .PP -\fBC1%, C2%, C3%\fP The residency percentage that Linux requested C1, C2, C3.... The system summary is the average of all CPUs in the system. Note that these are software, reflecting what was requested. The hardware counters reflect what was actually achieved. +\fBC1%, C2%, C3%\fP The residency percentage that Linux requested C1, C2, C3.... The system summary is the average of all CPUs in the system. Note that these are software, reflecting what was requested. The hardware counters reflect what was actually achieved. These counters are in the "pct_idle" group, which is enabled by default. .PP -\fBCPU%c1, CPU%c3, CPU%c6, CPU%c7\fP show the percentage residency in hardware core idle states. These numbers are from hardware residency counters. +\fBCPU%c1, CPU%c3, CPU%c6, CPU%c7\fP show the percentage residency in hardware core idle states. These numbers are from hardware residency counters and are in the "hwidle" group, which is enabled, by default. .PP \fBCoreTmp\fP Degrees Celsius reported by the per-core Digital Thermal Sensor. .PP diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index df0391bedcde..ab184f95cdaf 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -153,7 +153,7 @@ struct msr_counter bic[] = { { 0x0, "TSC_MHz", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "IRQ", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "SMI", NULL, 32, 0, FORMAT_DELTA, NULL, 0 }, - { 0x0, "sysfs", NULL, 0, 0, 0, NULL, 0 }, + { 0x0, "cpuidle", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "CPU%c1", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "CPU%c3", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "CPU%c6", NULL, 0, 0, 0, NULL, 0 }, @@ -206,6 +206,7 @@ struct msr_counter bic[] = { { 0x0, "Sys_J", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "NMI", NULL, 0, 0, 0, NULL, 0 }, { 0x0, "CPU%c1e", NULL, 0, 0, 0, NULL, 0 }, + { 0x0, "pct_idle", NULL, 0, 0, 0, NULL, 0 }, }; #define MAX_BIC (sizeof(bic) / sizeof(struct msr_counter)) @@ -219,7 +220,7 @@ struct msr_counter bic[] = { #define BIC_TSC_MHz (1ULL << 7) #define BIC_IRQ (1ULL << 8) #define BIC_SMI (1ULL << 9) -#define BIC_sysfs (1ULL << 10) +#define BIC_cpuidle (1ULL << 10) #define BIC_CPU_c1 (1ULL << 11) #define BIC_CPU_c3 (1ULL << 12) #define BIC_CPU_c6 (1ULL << 13) @@ -272,17 +273,20 @@ struct msr_counter bic[] = { #define BIC_Sys_J (1ULL << 60) #define BIC_NMI (1ULL << 61) #define BIC_CPU_c1e (1ULL << 62) +#define BIC_pct_idle (1ULL << 63) #define BIC_GROUP_TOPOLOGY (BIC_Package | BIC_Node | BIC_CoreCnt | BIC_PkgCnt | BIC_Core | BIC_CPU | BIC_Die) #define BIC_GROUP_THERMAL_PWR (BIC_CoreTmp | BIC_PkgTmp | BIC_PkgWatt | BIC_CorWatt | BIC_GFXWatt | BIC_RAMWatt | BIC_PKG__ | BIC_RAM__ | BIC_SysWatt) #define BIC_GROUP_FREQUENCY (BIC_Avg_MHz | BIC_Busy | BIC_Bzy_MHz | BIC_TSC_MHz | BIC_GFXMHz | BIC_GFXACTMHz | BIC_SAMMHz | BIC_SAMACTMHz | BIC_UNCORE_MHZ) -#define BIC_GROUP_IDLE (BIC_Busy | BIC_sysfs | BIC_CPU_c1 | BIC_CPU_c3 | BIC_CPU_c6 | BIC_CPU_c7 | BIC_GFX_rc6 | BIC_Pkgpc2 | BIC_Pkgpc3 | BIC_Pkgpc6 | BIC_Pkgpc7 | BIC_Pkgpc8 | BIC_Pkgpc9 | BIC_Pkgpc10 | BIC_CPU_LPI | BIC_SYS_LPI | BIC_Mod_c6 | BIC_Totl_c0 | BIC_Any_c0 | BIC_GFX_c0 | BIC_CPUGFX | BIC_SAM_mc6 | BIC_Diec6) +#define BIC_GROUP_HW_IDLE (BIC_Busy | BIC_CPU_c1 | BIC_CPU_c3 | BIC_CPU_c6 | BIC_CPU_c7 | BIC_GFX_rc6 | BIC_Pkgpc2 | BIC_Pkgpc3 | BIC_Pkgpc6 | BIC_Pkgpc7 | BIC_Pkgpc8 | BIC_Pkgpc9 | BIC_Pkgpc10 | BIC_CPU_LPI | BIC_SYS_LPI | BIC_Mod_c6 | BIC_Totl_c0 | BIC_Any_c0 | BIC_GFX_c0 | BIC_CPUGFX | BIC_SAM_mc6 | BIC_Diec6) +#define BIC_GROUP_SW_IDLE (BIC_Busy | BIC_cpuidle | BIC_pct_idle ) +#define BIC_GROUP_IDLE (BIC_GROUP_HW_IDLE | BIC_pct_idle) #define BIC_OTHER (BIC_IRQ | BIC_NMI | BIC_SMI | BIC_ThreadC | BIC_CoreTmp | BIC_IPC) -#define BIC_DISABLED_BY_DEFAULT (BIC_USEC | BIC_TOD | BIC_APIC | BIC_X2APIC) +#define BIC_DISABLED_BY_DEFAULT (BIC_USEC | BIC_TOD | BIC_APIC | BIC_X2APIC | BIC_cpuidle) unsigned long long bic_enabled = (0xFFFFFFFFFFFFFFFFULL & ~BIC_DISABLED_BY_DEFAULT); -unsigned long long bic_present = BIC_USEC | BIC_TOD | BIC_sysfs | BIC_APIC | BIC_X2APIC; +unsigned long long bic_present = BIC_USEC | BIC_TOD | BIC_cpuidle | BIC_pct_idle | BIC_APIC | BIC_X2APIC; #define DO_BIC(COUNTER_NAME) (bic_enabled & bic_present & COUNTER_NAME) #define DO_BIC_READ(COUNTER_NAME) (bic_present & COUNTER_NAME) @@ -2362,6 +2366,15 @@ unsigned long long bic_lookup(char *name_list, enum show_hide_mode mode) } else if (!strcmp(name_list, "idle")) { retval |= BIC_GROUP_IDLE; break; + } else if (!strcmp(name_list, "swidle")) { + retval |= BIC_GROUP_SW_IDLE; + break; + } else if (!strcmp(name_list, "sysfs")) { /* legacy compatibility */ + retval |= BIC_GROUP_SW_IDLE; + break; + } else if (!strcmp(name_list, "hwidle")) { + retval |= BIC_GROUP_HW_IDLE; + break; } else if (!strcmp(name_list, "frequency")) { retval |= BIC_GROUP_FREQUENCY; break; @@ -2372,6 +2385,7 @@ unsigned long long bic_lookup(char *name_list, enum show_hide_mode mode) } if (i == MAX_BIC) { + fprintf(stderr, "deferred %s\n", name_list); if (mode == SHOW_LIST) { deferred_add_names[deferred_add_index++] = name_list; if (deferred_add_index >= MAX_DEFERRED) { @@ -10269,6 +10283,9 @@ void probe_cpuidle_residency(void) int min_state = 1024, max_state = 0; char *sp; + if (!DO_BIC(BIC_pct_idle)) + return; + for (state = 10; state >= 0; --state) { sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", base_cpu, state); @@ -10291,7 +10308,7 @@ void probe_cpuidle_residency(void) sprintf(path, "cpuidle/state%d/time", state); - if (!DO_BIC(BIC_sysfs) && !is_deferred_add(name_buf)) + if (!DO_BIC(BIC_pct_idle) && !is_deferred_add(name_buf)) continue; if (is_deferred_skip(name_buf)) @@ -10315,6 +10332,9 @@ void probe_cpuidle_counts(void) int min_state = 1024, max_state = 0; char *sp; + if (!DO_BIC(BIC_cpuidle)) + return; + for (state = 10; state >= 0; --state) { sprintf(path, "/sys/devices/system/cpu/cpu%d/cpuidle/state%d/name", base_cpu, state); @@ -10327,7 +10347,7 @@ void probe_cpuidle_counts(void) remove_underbar(name_buf); - if (!DO_BIC(BIC_sysfs) && !is_deferred_add(name_buf)) + if (!DO_BIC(BIC_cpuidle) && !is_deferred_add(name_buf)) continue; if (is_deferred_skip(name_buf)) -- 2.51.0 From 03e00e373cab981ad808271b2650700cfa0fbda6 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Sun, 6 Apr 2025 14:49:20 -0400 Subject: [PATCH 06/16] tools/power turbostat: v2025.05.06 Support up to 8192 processors Add cpuidle governor debug telemetry, disabled by default Update default output to exclude cpuidle invocation counts Bug fixes Signed-off-by: Len Brown --- tools/power/x86/turbostat/turbostat.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c index ab184f95cdaf..1b9fdc1a7ee8 100644 --- a/tools/power/x86/turbostat/turbostat.c +++ b/tools/power/x86/turbostat/turbostat.c @@ -9594,7 +9594,7 @@ int get_and_dump_counters(void) void print_version() { - fprintf(outf, "turbostat version 2025.02.02 - Len Brown \n"); + fprintf(outf, "turbostat version 2025.04.06 - Len Brown \n"); } #define COMMAND_LINE_SIZE 2048 -- 2.51.0 From 0efdedb3358aa78102967f242379686f94315830 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Thomas=20Wei=C3=9Fschuh?= Date: Wed, 2 Apr 2025 21:21:57 +0100 Subject: [PATCH 07/16] tools/include: make uapi/linux/types.h usable from assembly MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit The "real" linux/types.h UAPI header gracefully degrades to a NOOP when included from assembly code. Mirror this behaviour in the tools/ variant. Test for __ASSEMBLER__ over __ASSEMBLY__ as the former is provided by the toolchain automatically. Reported-by: Mark Brown Closes: https://lore.kernel.org/lkml/af553c62-ca2f-4956-932c-dd6e3a126f58@sirena.org.uk/ Fixes: c9fbaa879508 ("selftests: vDSO: parse_vdso: Use UAPI headers instead of libc headers") Signed-off-by: Thomas Weißschuh Link: https://patch.msgid.link/20250321-uapi-consistency-v1-1-439070118dc0@linutronix.de Signed-off-by: Mark Brown Reviewed-by: Mark Brown Signed-off-by: Linus Torvalds --- tools/include/uapi/linux/types.h | 3 +++ 1 file changed, 3 insertions(+) diff --git a/tools/include/uapi/linux/types.h b/tools/include/uapi/linux/types.h index 91fa51a9c31d..85aa327245c6 100644 --- a/tools/include/uapi/linux/types.h +++ b/tools/include/uapi/linux/types.h @@ -4,6 +4,8 @@ #include +#ifndef __ASSEMBLER__ + /* copied from linux:include/uapi/linux/types.h */ #define __bitwise typedef __u16 __bitwise __le16; @@ -20,4 +22,5 @@ typedef __u32 __bitwise __wsum; #define __aligned_be64 __be64 __attribute__((aligned(8))) #define __aligned_le64 __le64 __attribute__((aligned(8))) +#endif /* __ASSEMBLER__ */ #endif /* _UAPI_LINUX_TYPES_H */ -- 2.51.0 From 0af2f6be1b4281385b618cb86ad946eded089ac8 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 6 Apr 2025 13:11:33 -0700 Subject: [PATCH 08/16] Linux 6.15-rc1 --- Makefile | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Makefile b/Makefile index e55726a71d95..38689a0c3605 100644 --- a/Makefile +++ b/Makefile @@ -1,8 +1,8 @@ # SPDX-License-Identifier: GPL-2.0 VERSION = 6 -PATCHLEVEL = 14 +PATCHLEVEL = 15 SUBLEVEL = 0 -EXTRAVERSION = +EXTRAVERSION = -rc1 NAME = Baby Opossum Posse # *DOCUMENTATION* -- 2.51.0 From b40c54648158f63557ae3bb594802a199632afaf Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Fri, 28 Mar 2025 12:32:59 +0100 Subject: [PATCH 09/16] xenbus: add module description Modules without a description now cause a warning: WARNING: modpost: missing MODULE_DESCRIPTION() in drivers/xen/xenbus/xenbus_probe_frontend.o Signed-off-by: Arnd Bergmann Reviewed-by: Juergen Gross Signed-off-by: Juergen Gross Message-ID: <20250328113302.2632353-1-arnd@kernel.org> --- drivers/xen/xenbus/xenbus_probe_frontend.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/xen/xenbus/xenbus_probe_frontend.c b/drivers/xen/xenbus/xenbus_probe_frontend.c index fcb335bb7b18..6d1819269cbe 100644 --- a/drivers/xen/xenbus/xenbus_probe_frontend.c +++ b/drivers/xen/xenbus/xenbus_probe_frontend.c @@ -513,4 +513,5 @@ static int __init boot_wait_for_devices(void) late_initcall(boot_wait_for_devices); #endif +MODULE_DESCRIPTION("Xen PV-device frontend support"); MODULE_LICENSE("GPL"); -- 2.51.0 From 0f2946bb172632e122d4033e0b03f85230a29510 Mon Sep 17 00:00:00 2001 From: Jason Andryuk Date: Mon, 31 Mar 2025 13:29:12 -0400 Subject: [PATCH 10/16] xen: Change xen-acpi-processor dom0 dependency xen-acpi-processor functions under a PVH dom0 with only a xen_initial_domain() runtime check. Change the Kconfig dependency from PV dom0 to generic dom0 to reflect that. Suggested-by: Jan Beulich Signed-off-by: Jason Andryuk Reviewed-by: Juergen Gross Tested-by: Jan Beulich Signed-off-by: Juergen Gross Message-ID: <20250331172913.51240-1-jason.andryuk@amd.com> --- drivers/xen/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/xen/Kconfig b/drivers/xen/Kconfig index f7d6f47971fd..24f485827e03 100644 --- a/drivers/xen/Kconfig +++ b/drivers/xen/Kconfig @@ -278,7 +278,7 @@ config XEN_PRIVCMD_EVENTFD config XEN_ACPI_PROCESSOR tristate "Xen ACPI processor" - depends on XEN && XEN_PV_DOM0 && X86 && ACPI_PROCESSOR && CPU_FREQ + depends on XEN && XEN_DOM0 && X86 && ACPI_PROCESSOR && CPU_FREQ default m help This ACPI processor uploads Power Management information to the Xen -- 2.51.0 From 87af633689ce16ddb166c80f32b120e50b1295de Mon Sep 17 00:00:00 2001 From: Roger Pau Monne Date: Mon, 7 Apr 2025 10:28:37 +0200 Subject: [PATCH 11/16] x86/xen: fix balloon target initialization for PVH dom0 MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit PVH dom0 re-uses logic from PV dom0, in which RAM ranges not assigned to dom0 are re-used as scratch memory to map foreign and grant pages. Such logic relies on reporting those unpopulated ranges as RAM to Linux, and mark them as reserved. This way Linux creates the underlying page structures required for metadata management. Such approach works fine on PV because the initial balloon target is calculated using specific Xen data, that doesn't take into account the memory type changes described above. However on HVM and PVH the initial balloon target is calculated using get_num_physpages(), and that function does take into account the unpopulated RAM regions used as scratch space for remote domain mappings. This leads to PVH dom0 having an incorrect initial balloon target, which causes malfunction (excessive memory freeing) of the balloon driver if the dom0 memory target is later adjusted from the toolstack. Fix this by using xen_released_pages to account for any pages that are part of the memory map, but are already unpopulated when the balloon driver is initialized. This accounts for any regions used for scratch remote mappings. Note on x86 xen_released_pages definition is moved to enlighten.c so it's uniformly available for all Xen-enabled builds. Take the opportunity to unify PV with PVH/HVM guests regarding the usage of get_num_physpages(), as that avoids having to add different logic for PV vs PVH in both balloon_add_regions() and arch_xen_unpopulated_init(). Much like a6aa4eb994ee, the code in this changeset should have been part of 38620fc4e893. Fixes: a6aa4eb994ee ('xen/x86: add extra pages to unpopulated-alloc if available') Signed-off-by: Roger Pau Monné Reviewed-by: Juergen Gross Cc: stable@vger.kernel.org Signed-off-by: Juergen Gross Message-ID: <20250407082838.65495-1-roger.pau@citrix.com> --- arch/x86/xen/enlighten.c | 10 ++++++++++ arch/x86/xen/setup.c | 3 --- drivers/xen/balloon.c | 34 ++++++++++++++++++++++++---------- 3 files changed, 34 insertions(+), 13 deletions(-) diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c index 43dcd8c7badc..1b7710bd0d05 100644 --- a/arch/x86/xen/enlighten.c +++ b/arch/x86/xen/enlighten.c @@ -70,6 +70,9 @@ EXPORT_SYMBOL(xen_start_flags); */ struct shared_info *HYPERVISOR_shared_info = &xen_dummy_shared_info; +/* Number of pages released from the initial allocation. */ +unsigned long xen_released_pages; + static __ref void xen_get_vendor(void) { init_cpu_devs(); @@ -466,6 +469,13 @@ int __init arch_xen_unpopulated_init(struct resource **res) xen_free_unpopulated_pages(1, &pg); } + /* + * Account for the region being in the physmap but unpopulated. + * The value in xen_released_pages is used by the balloon + * driver to know how much of the physmap is unpopulated and + * set an accurate initial memory target. + */ + xen_released_pages += xen_extra_mem[i].n_pfns; /* Zero so region is not also added to the balloon driver. */ xen_extra_mem[i].n_pfns = 0; } diff --git a/arch/x86/xen/setup.c b/arch/x86/xen/setup.c index c3db71d96c43..3823e52aef52 100644 --- a/arch/x86/xen/setup.c +++ b/arch/x86/xen/setup.c @@ -37,9 +37,6 @@ #define GB(x) ((uint64_t)(x) * 1024 * 1024 * 1024) -/* Number of pages released from the initial allocation. */ -unsigned long xen_released_pages; - /* Memory map would allow PCI passthrough. */ bool xen_pv_pci_possible; diff --git a/drivers/xen/balloon.c b/drivers/xen/balloon.c index 65d4e7fa1eb8..8c852807ba1c 100644 --- a/drivers/xen/balloon.c +++ b/drivers/xen/balloon.c @@ -679,7 +679,7 @@ void xen_free_ballooned_pages(unsigned int nr_pages, struct page **pages) } EXPORT_SYMBOL(xen_free_ballooned_pages); -static void __init balloon_add_regions(void) +static int __init balloon_add_regions(void) { unsigned long start_pfn, pages; unsigned long pfn, extra_pfn_end; @@ -702,26 +702,38 @@ static void __init balloon_add_regions(void) for (pfn = start_pfn; pfn < extra_pfn_end; pfn++) balloon_append(pfn_to_page(pfn)); - balloon_stats.total_pages += extra_pfn_end - start_pfn; + /* + * Extra regions are accounted for in the physmap, but need + * decreasing from current_pages to balloon down the initial + * allocation, because they are already accounted for in + * total_pages. + */ + if (extra_pfn_end - start_pfn >= balloon_stats.current_pages) { + WARN(1, "Extra pages underflow current target"); + return -ERANGE; + } + balloon_stats.current_pages -= extra_pfn_end - start_pfn; } + + return 0; } static int __init balloon_init(void) { struct task_struct *task; + int rc; if (!xen_domain()) return -ENODEV; pr_info("Initialising balloon driver\n"); -#ifdef CONFIG_XEN_PV - balloon_stats.current_pages = xen_pv_domain() - ? min(xen_start_info->nr_pages - xen_released_pages, max_pfn) - : get_num_physpages(); -#else - balloon_stats.current_pages = get_num_physpages(); -#endif + if (xen_released_pages >= get_num_physpages()) { + WARN(1, "Released pages underflow current target"); + return -ERANGE; + } + + balloon_stats.current_pages = get_num_physpages() - xen_released_pages; balloon_stats.target_pages = balloon_stats.current_pages; balloon_stats.balloon_low = 0; balloon_stats.balloon_high = 0; @@ -738,7 +750,9 @@ static int __init balloon_init(void) register_sysctl_init("xen/balloon", balloon_table); #endif - balloon_add_regions(); + rc = balloon_add_regions(); + if (rc) + return rc; task = kthread_run(balloon_thread, NULL, "xen-balloon"); if (IS_ERR(task)) { -- 2.51.0 From 64a66e2c3b3113dc78a6124e14825d68ddc2e188 Mon Sep 17 00:00:00 2001 From: Roger Pau Monne Date: Mon, 7 Apr 2025 12:18:41 +0200 Subject: [PATCH 12/16] x86/xen: disable CPU idle and frequency drivers for PVH dom0 MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit When running as a PVH dom0 the ACPI tables exposed to Linux are (mostly) the native ones, thus exposing the C and P states, that can lead to attachment of CPU idle and frequency drivers. However the entity in control of the CPU C and P states is Xen, as dom0 doesn't have a full view of the system load, neither has all CPUs assigned and identity pinned. Like it's done for classic PV guests, prevent Linux from using idle or frequency state drivers when running as a PVH dom0. On an AMD EPYC 7543P system without this fix a Linux PVH dom0 will keep the host CPUs spinning at 100% even when dom0 is completely idle, as it's attempting to use the acpi_idle driver. Signed-off-by: Roger Pau Monné Reviewed-by: Jason Andryuk Signed-off-by: Juergen Gross Message-ID: <20250407101842.67228-1-roger.pau@citrix.com> --- arch/x86/xen/enlighten_pvh.c | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-) diff --git a/arch/x86/xen/enlighten_pvh.c b/arch/x86/xen/enlighten_pvh.c index 0e3d930bcb89..9d25d9373945 100644 --- a/arch/x86/xen/enlighten_pvh.c +++ b/arch/x86/xen/enlighten_pvh.c @@ -1,5 +1,7 @@ // SPDX-License-Identifier: GPL-2.0 #include +#include +#include #include #include @@ -123,8 +125,23 @@ static void __init pvh_arch_setup(void) { pvh_reserve_extra_memory(); - if (xen_initial_domain()) + if (xen_initial_domain()) { xen_add_preferred_consoles(); + + /* + * Disable usage of CPU idle and frequency drivers: when + * running as hardware domain the exposed native ACPI tables + * causes idle and/or frequency drivers to attach and + * malfunction. It's Xen the entity that controls the idle and + * frequency states. + * + * For unprivileged domains the exposed ACPI tables are + * fabricated and don't contain such data. + */ + disable_cpuidle(); + disable_cpufreq(); + WARN_ON(xen_set_default_idle()); + } } void __init xen_pvh_init(struct boot_params *boot_params) -- 2.51.0 From 715ad3e0ec2b13c27335749f27a5c9f0c0e84064 Mon Sep 17 00:00:00 2001 From: Juergen Gross Date: Thu, 27 Mar 2025 20:06:02 +0100 Subject: [PATCH 13/16] xen: fix multicall debug feature Initializing a percpu variable with the address of a struct tagged as .initdata is breaking the build with CONFIG_SECTION_MISMATCH_WARN_ONLY not set to "y". Fix that by using an access function instead returning the .initdata struct address if the percpu space of the struct hasn't been allocated yet. Fixes: 368990a7fe30 ("xen: fix multicall debug data referencing") Reported-by: Borislav Petkov Reviewed-by: Boris Ostrovsky Acked-by: "Borislav Petkov (AMD)" Tested-by: "Borislav Petkov (AMD)" Signed-off-by: Juergen Gross Message-ID: <20250327190602.26015-1-jgross@suse.com> --- arch/x86/xen/multicalls.c | 26 ++++++++++++++------------ arch/x86/xen/smp_pv.c | 1 - arch/x86/xen/xen-ops.h | 3 --- 3 files changed, 14 insertions(+), 16 deletions(-) diff --git a/arch/x86/xen/multicalls.c b/arch/x86/xen/multicalls.c index 10c660fae8b3..7237d56a9d3f 100644 --- a/arch/x86/xen/multicalls.c +++ b/arch/x86/xen/multicalls.c @@ -54,14 +54,20 @@ struct mc_debug_data { static DEFINE_PER_CPU(struct mc_buffer, mc_buffer); static struct mc_debug_data mc_debug_data_early __initdata; -static DEFINE_PER_CPU(struct mc_debug_data *, mc_debug_data) = - &mc_debug_data_early; static struct mc_debug_data __percpu *mc_debug_data_ptr; DEFINE_PER_CPU(unsigned long, xen_mc_irq_flags); static struct static_key mc_debug __ro_after_init; static bool mc_debug_enabled __initdata; +static struct mc_debug_data * __ref get_mc_debug(void) +{ + if (!mc_debug_data_ptr) + return &mc_debug_data_early; + + return this_cpu_ptr(mc_debug_data_ptr); +} + static int __init xen_parse_mc_debug(char *arg) { mc_debug_enabled = true; @@ -71,20 +77,16 @@ static int __init xen_parse_mc_debug(char *arg) } early_param("xen_mc_debug", xen_parse_mc_debug); -void mc_percpu_init(unsigned int cpu) -{ - per_cpu(mc_debug_data, cpu) = per_cpu_ptr(mc_debug_data_ptr, cpu); -} - static int __init mc_debug_enable(void) { unsigned long flags; + struct mc_debug_data __percpu *mcdb; if (!mc_debug_enabled) return 0; - mc_debug_data_ptr = alloc_percpu(struct mc_debug_data); - if (!mc_debug_data_ptr) { + mcdb = alloc_percpu(struct mc_debug_data); + if (!mcdb) { pr_err("xen_mc_debug inactive\n"); static_key_slow_dec(&mc_debug); return -ENOMEM; @@ -93,7 +95,7 @@ static int __init mc_debug_enable(void) /* Be careful when switching to percpu debug data. */ local_irq_save(flags); xen_mc_flush(); - mc_percpu_init(0); + mc_debug_data_ptr = mcdb; local_irq_restore(flags); pr_info("xen_mc_debug active\n"); @@ -155,7 +157,7 @@ void xen_mc_flush(void) trace_xen_mc_flush(b->mcidx, b->argidx, b->cbidx); if (static_key_false(&mc_debug)) { - mcdb = __this_cpu_read(mc_debug_data); + mcdb = get_mc_debug(); memcpy(mcdb->entries, b->entries, b->mcidx * sizeof(struct multicall_entry)); } @@ -235,7 +237,7 @@ struct multicall_space __xen_mc_entry(size_t args) ret.mc = &b->entries[b->mcidx]; if (static_key_false(&mc_debug)) { - struct mc_debug_data *mcdb = __this_cpu_read(mc_debug_data); + struct mc_debug_data *mcdb = get_mc_debug(); mcdb->caller[b->mcidx] = __builtin_return_address(0); mcdb->argsz[b->mcidx] = args; diff --git a/arch/x86/xen/smp_pv.c b/arch/x86/xen/smp_pv.c index 688ff59318ae..9bb8ff8bff30 100644 --- a/arch/x86/xen/smp_pv.c +++ b/arch/x86/xen/smp_pv.c @@ -305,7 +305,6 @@ static int xen_pv_kick_ap(unsigned int cpu, struct task_struct *idle) return rc; xen_pmu_init(cpu); - mc_percpu_init(cpu); /* * Why is this a BUG? If the hypercall fails then everything can be diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h index 63c13a2ccf55..25e318ef27d6 100644 --- a/arch/x86/xen/xen-ops.h +++ b/arch/x86/xen/xen-ops.h @@ -261,9 +261,6 @@ void xen_mc_callback(void (*fn)(void *), void *data); */ struct multicall_space xen_mc_extend_args(unsigned long op, size_t arg_size); -/* Do percpu data initialization for multicalls. */ -void mc_percpu_init(unsigned int cpu); - extern bool is_xen_pmu; irqreturn_t xen_pmu_irq_handler(int irq, void *dev_id); -- 2.51.0 From cd9c058489053e172a6654cad82ee936d1b09fab Mon Sep 17 00:00:00 2001 From: John Ernberg Date: Fri, 2 May 2025 11:40:55 +0000 Subject: [PATCH 14/16] xen: swiotlb: Use swiotlb bouncing if kmalloc allocation demands it Xen swiotlb support was missed when the patch set starting with 4ab5f8ec7d71 ("mm/slab: decouple ARCH_KMALLOC_MINALIGN from ARCH_DMA_MINALIGN") was merged. When running Xen on iMX8QXP, a SoC without IOMMU, the effect was that USB transfers ended up corrupted when there was more than one URB inflight at the same time. Add a call to dma_kmalloc_needs_bounce() to make sure that allocations too small for DMA get bounced via swiotlb. Closes: https://lore.kernel.org/linux-usb/ab2776f0-b838-4cf6-a12a-c208eb6aad59@actia.se/ Fixes: 4ab5f8ec7d71 ("mm/slab: decouple ARCH_KMALLOC_MINALIGN from ARCH_DMA_MINALIGN") Cc: stable@kernel.org # v6.5+ Signed-off-by: John Ernberg Reviewed-by: Stefano Stabellini Signed-off-by: Juergen Gross Message-ID: <20250502114043.1968976-2-john.ernberg@actia.se> --- drivers/xen/swiotlb-xen.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c index 1f65795cf5d7..ef56a2500ed6 100644 --- a/drivers/xen/swiotlb-xen.c +++ b/drivers/xen/swiotlb-xen.c @@ -217,6 +217,7 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page, * buffering it. */ if (dma_capable(dev, dev_addr, size, true) && + !dma_kmalloc_needs_bounce(dev, size, dir) && !range_straddles_page_boundary(phys, size) && !xen_arch_need_swiotlb(dev, phys, dev_addr) && !is_swiotlb_force_bounce(dev)) -- 2.51.0 From 90989869baae47ee2aa3bcb6f6eb9fbbe4287958 Mon Sep 17 00:00:00 2001 From: Jason Andryuk Date: Tue, 6 May 2025 16:44:56 -0400 Subject: [PATCH 15/16] xenbus: Allow PVH dom0 a non-local xenstore Make xenbus_init() allow a non-local xenstore for a PVH dom0 - it is currently forced to XS_LOCAL. With Hyperlaunch booting dom0 and a xenstore stubdom, dom0 can be handled as a regular XS_HVM following the late init path. Ideally we'd drop the use of xen_initial_domain() and just check for the event channel instead. However, ARM has a xen,enhanced no-xenstore mode, where the event channel and PFN would both be 0. Retain the xen_initial_domain() check, and use that for an additional check when the event channel is 0. Check the full 64bit HVM_PARAM_STORE_EVTCHN value to catch the off chance that high bits are set for the 32bit event channel. Signed-off-by: Jason Andryuk Change-Id: I5506da42e4c6b8e85079fefb2f193c8de17c7437 Reviewed-by: Stefano Stabellini Signed-off-by: Juergen Gross Message-ID: <20250506204456.5220-1-jason.andryuk@amd.com> --- drivers/xen/xenbus/xenbus_probe.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c index 6d32ffb01136..86fe6e779056 100644 --- a/drivers/xen/xenbus/xenbus_probe.c +++ b/drivers/xen/xenbus/xenbus_probe.c @@ -966,9 +966,15 @@ static int __init xenbus_init(void) if (xen_pv_domain()) xen_store_domain_type = XS_PV; if (xen_hvm_domain()) + { xen_store_domain_type = XS_HVM; - if (xen_hvm_domain() && xen_initial_domain()) - xen_store_domain_type = XS_LOCAL; + err = hvm_get_parameter(HVM_PARAM_STORE_EVTCHN, &v); + if (err) + goto out_error; + xen_store_evtchn = (int)v; + if (!v && xen_initial_domain()) + xen_store_domain_type = XS_LOCAL; + } if (xen_pv_domain() && !xen_start_info->store_evtchn) xen_store_domain_type = XS_LOCAL; if (xen_pv_domain() && xen_start_info->store_evtchn) @@ -987,10 +993,6 @@ static int __init xenbus_init(void) xen_store_interface = gfn_to_virt(xen_store_gfn); break; case XS_HVM: - err = hvm_get_parameter(HVM_PARAM_STORE_EVTCHN, &v); - if (err) - goto out_error; - xen_store_evtchn = (int)v; err = hvm_get_parameter(HVM_PARAM_STORE_PFN, &v); if (err) goto out_error; -- 2.51.0 From 1f0304dfd9d217c2f8b04a9ef4b3258a66eedd27 Mon Sep 17 00:00:00 2001 From: Jason Andryuk Date: Tue, 6 May 2025 17:09:33 -0400 Subject: [PATCH 16/16] xenbus: Use kref to track req lifetime MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit Marek reported seeing a NULL pointer fault in the xenbus_thread callstack: BUG: kernel NULL pointer dereference, address: 0000000000000000 RIP: e030:__wake_up_common+0x4c/0x180 Call Trace: __wake_up_common_lock+0x82/0xd0 process_msg+0x18e/0x2f0 xenbus_thread+0x165/0x1c0 process_msg+0x18e is req->cb(req). req->cb is set to xs_wake_up(), a thin wrapper around wake_up(), or xenbus_dev_queue_reply(). It seems like it was xs_wake_up() in this case. It seems like req may have woken up the xs_wait_for_reply(), which kfree()ed the req. When xenbus_thread resumes, it faults on the zero-ed data. Linux Device Drivers 2nd edition states: "Normally, a wake_up call can cause an immediate reschedule to happen, meaning that other processes might run before wake_up returns." ... which would match the behaviour observed. Change to keeping two krefs on each request. One for the caller, and one for xenbus_thread. Each will kref_put() when finished, and the last will free it. This use of kref matches the description in Documentation/core-api/kref.rst Link: https://lore.kernel.org/xen-devel/ZO0WrR5J0xuwDIxW@mail-itl/ Reported-by: Marek Marczykowski-Górecki Fixes: fd8aa9095a95 ("xen: optimize xenbus driver for multiple concurrent xenstore accesses") Cc: stable@vger.kernel.org Signed-off-by: Jason Andryuk Reviewed-by: Juergen Gross Signed-off-by: Juergen Gross Message-ID: <20250506210935.5607-1-jason.andryuk@amd.com> --- drivers/xen/xenbus/xenbus.h | 2 ++ drivers/xen/xenbus/xenbus_comms.c | 9 ++++----- drivers/xen/xenbus/xenbus_dev_frontend.c | 2 +- drivers/xen/xenbus/xenbus_xs.c | 18 ++++++++++++++++-- 4 files changed, 23 insertions(+), 8 deletions(-) diff --git a/drivers/xen/xenbus/xenbus.h b/drivers/xen/xenbus/xenbus.h index 13821e7e825e..9ac0427724a3 100644 --- a/drivers/xen/xenbus/xenbus.h +++ b/drivers/xen/xenbus/xenbus.h @@ -77,6 +77,7 @@ enum xb_req_state { struct xb_req_data { struct list_head list; wait_queue_head_t wq; + struct kref kref; struct xsd_sockmsg msg; uint32_t caller_req_id; enum xsd_sockmsg_type type; @@ -103,6 +104,7 @@ int xb_init_comms(void); void xb_deinit_comms(void); int xs_watch_msg(struct xs_watch_event *event); void xs_request_exit(struct xb_req_data *req); +void xs_free_req(struct kref *kref); int xenbus_match(struct device *_dev, const struct device_driver *_drv); int xenbus_dev_probe(struct device *_dev); diff --git a/drivers/xen/xenbus/xenbus_comms.c b/drivers/xen/xenbus/xenbus_comms.c index e5fda0256feb..82df2da1b880 100644 --- a/drivers/xen/xenbus/xenbus_comms.c +++ b/drivers/xen/xenbus/xenbus_comms.c @@ -309,8 +309,8 @@ static int process_msg(void) virt_wmb(); req->state = xb_req_state_got_reply; req->cb(req); - } else - kfree(req); + } + kref_put(&req->kref, xs_free_req); } mutex_unlock(&xs_response_mutex); @@ -386,14 +386,13 @@ static int process_writes(void) state.req->msg.type = XS_ERROR; state.req->err = err; list_del(&state.req->list); - if (state.req->state == xb_req_state_aborted) - kfree(state.req); - else { + if (state.req->state != xb_req_state_aborted) { /* write err, then update state */ virt_wmb(); state.req->state = xb_req_state_got_reply; wake_up(&state.req->wq); } + kref_put(&state.req->kref, xs_free_req); mutex_unlock(&xb_write_mutex); diff --git a/drivers/xen/xenbus/xenbus_dev_frontend.c b/drivers/xen/xenbus/xenbus_dev_frontend.c index 46f8916597e5..f5c21ba64df5 100644 --- a/drivers/xen/xenbus/xenbus_dev_frontend.c +++ b/drivers/xen/xenbus/xenbus_dev_frontend.c @@ -406,7 +406,7 @@ void xenbus_dev_queue_reply(struct xb_req_data *req) mutex_unlock(&u->reply_mutex); kfree(req->body); - kfree(req); + kref_put(&req->kref, xs_free_req); kref_put(&u->kref, xenbus_file_free); diff --git a/drivers/xen/xenbus/xenbus_xs.c b/drivers/xen/xenbus/xenbus_xs.c index d32c726f7a12..dcf9182c8451 100644 --- a/drivers/xen/xenbus/xenbus_xs.c +++ b/drivers/xen/xenbus/xenbus_xs.c @@ -112,6 +112,12 @@ static void xs_suspend_exit(void) wake_up_all(&xs_state_enter_wq); } +void xs_free_req(struct kref *kref) +{ + struct xb_req_data *req = container_of(kref, struct xb_req_data, kref); + kfree(req); +} + static uint32_t xs_request_enter(struct xb_req_data *req) { uint32_t rq_id; @@ -237,6 +243,12 @@ static void xs_send(struct xb_req_data *req, struct xsd_sockmsg *msg) req->caller_req_id = req->msg.req_id; req->msg.req_id = xs_request_enter(req); + /* + * Take 2nd ref. One for this thread, and the second for the + * xenbus_thread. + */ + kref_get(&req->kref); + mutex_lock(&xb_write_mutex); list_add_tail(&req->list, &xb_write_list); notify = list_is_singular(&xb_write_list); @@ -261,8 +273,8 @@ static void *xs_wait_for_reply(struct xb_req_data *req, struct xsd_sockmsg *msg) if (req->state == xb_req_state_queued || req->state == xb_req_state_wait_reply) req->state = xb_req_state_aborted; - else - kfree(req); + + kref_put(&req->kref, xs_free_req); mutex_unlock(&xb_write_mutex); return ret; @@ -291,6 +303,7 @@ int xenbus_dev_request_and_reply(struct xsd_sockmsg *msg, void *par) req->cb = xenbus_dev_queue_reply; req->par = par; req->user_req = true; + kref_init(&req->kref); xs_send(req, msg); @@ -319,6 +332,7 @@ static void *xs_talkv(struct xenbus_transaction t, req->num_vecs = num_vecs; req->cb = xs_wake_up; req->user_req = false; + kref_init(&req->kref); msg.req_id = 0; msg.tx_id = t.id; -- 2.51.0