f388f60ca9: BUG:soft_lockup-CPU##stuck_for#s![swapper:#]
kernel test robot
build botHello,
by this commit, we notice big config diff [1]
then in this rcutorture tests, parent runs quite clean, f388f60ca9 shows various random issues.
=========================================================================================
tbox_group/testcase/rootfs/kconfig/compiler/runtime/test/torture_type:
vm-snb/rcutorture/debian-11.1-i386-20220923.cgz/i386-randconfig-r071-20250410/gcc-12/300s/default/tasks-tracing
fc2d5cbe541032e7 f388f60ca9041a95c9b3f157d31
---------------- ---------------------------
fail:runs %reproduction fail:runs
| | |
:500 30% 149:500 last_state.booting
:500 7% 35:500 dmesg.BUG:kernel_hang_in_boot_stage
:500 9% 45:500 dmesg.BUG:soft_lockup-CPU##stuck_for#s![swapper:#]
:500 10% 51:500 dmesg.BUG:workqueue_lockup-pool
:500 0% 1:500 dmesg.EIP:__timer_delete_sync
:500 1% 5:500 dmesg.EIP:_raw_spin_unlock_irq
:500 0% 2:500 dmesg.EIP:_raw_spin_unlock_irqrestore
:500 0% 1:500 dmesg.EIP:console_emit_next_record
:500 0% 1:500 dmesg.EIP:handle_softirqs
:500 1% 3:500 dmesg.EIP:lock_acquire
:500 0% 2:500 dmesg.EIP:lock_release
:500 0% 1:500 dmesg.EIP:queue_delayed_work_on
:500 9% 45:500 dmesg.EIP:timekeeping_notify
:500 3% 14:500 dmesg.INFO:rcu_preempt_detected_stalls_on_CPUs/tasks
:500 6% 32:500 dmesg.INFO:task_blocked_for_more_than#seconds
:500 1% 3:500 dmesg.IP-Config:Auto-configuration_of_network_failed
:500 9% 45:500 dmesg.Kernel_panic-not_syncing:softlockup:hung_tasks
:500 29% 146:500 dmesg.boot_failures
we don't have enough knowledge to dig deep these issues. so just make this report to consult with you if these issues are related with config diff. and if so, is this config diff reasonable by this commit?
below our normal report just FYI.
kernel test robot noticed "BUG:soft_lockup-CPU##stuck_for#s![swapper:#]" on:
commit: f388f60ca9041a95c9b3f157d316ed7c8f297e44 ("x86/cpu: Drop configuration options for early 64-bit CPUs")
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
[test failed on linus/master e618ee89561b6b0fdc69f79e6fd0c33375d3e6b4]
[test failed on linux-next/master 01c6df60d5d4ae00cd5c1648818744838bba7763]
in testcase: rcutorture
version:
with following parameters:
runtime: 300s test: default torture_type: tasks-tracing
config: i386-randconfig-r071-20250410
compiler: gcc-12
test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16G
(please refer to attached dmesg/kmsg for entire log/backtrace)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-lkp/[email protected]
[ 721.016745][ C0] watchdog: BUG: soft lockup - CPU#0 stuck for 626s! [swapper:1] [ 721.016779][ C0] CPU#0 Utilization every 96s during lockup: [ 721.016779][ C0] #1: 39% system, 0% softirq, 0% hardirq, 0% idle [ 721.016779][ C0] #2: 42% system, 0% softirq, 0% hardirq, 0% idle [ 721.016779][ C0] #3: 47% system, 0% softirq, 0% hardirq, 0% idle [ 721.016779][ C0] #4: 34% system, 0% softirq, 0% hardirq, 0% idle [ 721.016779][ C0] #5: 32% system, 0% softirq, 0% hardirq, 0% idle [ 721.016779][ C0] Modules linked in: [ 721.016779][ C0] irq event stamp: 159506 [ 721.016779][ C0] hardirqs last enabled at (159505): timekeeping_notify (arch/x86/include/asm/irqflags.h:42 arch/x86/include/asm/irqflags.h:97 arch/x86/include/asm/irqflags.h:155 include/linux/stop_machine.h:154 include/linux/stop_machine.h:161 kernel/time/timekeeping.c:1521) [ 721.016779][ C0] hardirqs last disabled at (159506): sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] softirqs last enabled at (159174): handle_softirqs (kernel/softirq.c:408 kernel/softirq.c:589) [ 721.016779][ C0] softirqs last disabled at (159159): __do_softirq (kernel/softirq.c:596) [ 721.016779][ C0] CPU: 0 UID: 0 PID: 1 Comm: swapper Not tainted 6.14.0-rc3-00037-gf388f60ca904 #1 [ 721.016779][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 [ 721.016779][ C0] EIP: timekeeping_notify (kernel/time/timekeeping.c:1522) [ 721.016779][ C0] Code: 5f e9 ff ff 8d 45 e8 e8 57 d4 ff ff 85 ff 74 16 8b 57 5c 85 d2 74 04 89 f8 ff d2 8b 87 88 00 00 00 e8 d5 3e ff ff 85 f6 75 9b <e8> 7f b9 00 00 31 c0 39 1d a4 70 14 84 0f 95 c0 f7 d8 8b 55 f0 2b All code ======== 0: 5f pop %rdi 1: e9 ff ff 8d 45 jmp 0x458e0005 6: e8 e8 57 d4 ff call 0xffffffffffd457f3 b: ff 85 ff 74 16 8b incl -0x74e98b01(%rbp) 11: 57 push %rdi 12: 5c pop %rsp 13: 85 d2 test %edx,%edx 15: 74 04 je 0x1b 17: 89 f8 mov %edi,%eax 19: ff d2 call *%rdx 1b: 8b 87 88 00 00 00 mov 0x88(%rdi),%eax 21: e8 d5 3e ff ff call 0xffffffffffff3efb 26: 85 f6 test %esi,%esi 28: 75 9b jne 0xffffffffffffffc5 2a:* e8 7f b9 00 00 call 0xb9ae <-- trapping instruction 2f: 31 c0 xor %eax,%eax 31: 39 1d a4 70 14 84 cmp %ebx,-0x7beb8f5c(%rip) # 0xffffffff841470db 37: 0f 95 c0 setne %al 3a: f7 d8 neg %eax 3c: 8b 55 f0 mov -0x10(%rbp),%edx 3f: 2b .byte 0x2b
Code starting with the faulting instruction
===========================================
0: e8 7f b9 00 00 call 0xb984 5: 31 c0 xor %eax,%eax 7: 39 1d a4 70 14 84 cmp %ebx,-0x7beb8f5c(%rip) # 0xffffffff841470b1 d: 0f 95 c0 setne %al 10: f7 d8 neg %eax 12: 8b 55 f0 mov -0x10(%rbp),%edx 15: 2b .byte 0x2b [ 721.016779][ C0] EAX: 00026f11 EBX: 8316b7e0 ECX: 00000006 EDX: 7e26f13f [ 721.016779][ C0] ESI: 00000200 EDI: 835e7220 EBP: 86d15ed8 ESP: 86d15ec0 [ 721.016779][ C0] DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068 EFLAGS: 00000206 [ 721.016779][ C0] CR0: 80050033 CR2: ffdaa000 CR3: 03a16000 CR4: 000406d0 [ 721.016779][ C0] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000 [ 721.016779][ C0] DR6: fffe0ff0 DR7: 00000400 [ 721.016779][ C0] Call Trace: [ 721.016779][ C0] ? show_regs (arch/x86/kernel/dumpstack.c:478) [ 721.016779][ C0] ? watchdog_timer_fn (kernel/watchdog.c:767) [ 721.016779][ C0] ? schedule_work (drivers/usb/core/hub.c:919) [ 721.016779][ C0] ? __hrtimer_run_queues+0x12f/0x1cf [ 721.016779][ C0] ? hrtimer_run_queues (kernel/time/hrtimer.c:2023) [ 721.016779][ C0] ? update_process_times (kernel/time/timer.c:2458 kernel/time/timer.c:2514) [ 721.016779][ C0] ? tick_periodic (kernel/time/tick-common.c:103) [ 721.016779][ C0] ? tick_handle_periodic (kernel/time/tick-common.c:144) [ 721.016779][ C0] ? vmware_sched_clock (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] ? __sysvec_apic_timer_interrupt (arch/x86/include/asm/trace/irq_vectors.h:41 arch/x86/include/asm/trace/irq_vectors.h:41 arch/x86/kernel/apic/apic.c:1056) [ 721.016779][ C0] ? sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049 arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] ? handle_exception (arch/x86/entry/entry_32.S:1055) [ 721.016779][ C0] ? vmware_sched_clock (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] ? timekeeping_notify (kernel/time/timekeeping.c:1522) [ 721.016779][ C0] ? vmware_sched_clock (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] ? timekeeping_notify (kernel/time/timekeeping.c:1522) [ 721.016779][ C0] __clocksource_select (kernel/time/clocksource.c:1077 (discriminator 1)) [ 721.016779][ C0] ? boot_override_clock (kernel/time/clocksource.c:1109) [ 721.016779][ C0] clocksource_select (kernel/time/clocksource.c:1094) [ 721.016779][ C0] clocksource_done_booting (kernel/time/clocksource.c:1118) [ 721.016779][ C0] do_one_initcall (init/main.c:1257) [ 721.016779][ C0] ? rdinit_setup (init/main.c:1305) [ 721.016779][ C0] do_initcalls (init/main.c:1318 init/main.c:1335) [ 721.016779][ C0] ? rest_init (init/main.c:1449) [ 721.016779][ C0] kernel_init_freeable (init/main.c:1572) [ 721.016779][ C0] kernel_init (init/main.c:1459) [ 721.016779][ C0] ret_from_fork (arch/x86/kernel/process.c:154) [ 721.016779][ C0] ? rest_init (init/main.c:1449) [ 721.016779][ C0] ret_from_fork_asm (arch/x86/entry/entry_32.S:737) [ 721.016779][ C0] entry_INT80_32 (arch/x86/entry/entry_32.S:945) [ 721.016779][ C0] Kernel panic - not syncing: softlockup: hung tasks [ 721.016779][ C0] CPU: 0 UID: 0 PID: 1 Comm: swapper Tainted: G L 6.14.0-rc3-00037-gf388f60ca904 #1 [ 721.016779][ C0] Tainted: [L]=SOFTLOCKUP [ 721.016779][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 [ 721.016779][ C0] Call Trace: [ 721.016779][ C0] dump_stack_lvl (lib/dump_stack.c:124) [ 721.016779][ C0] dump_stack (lib/dump_stack.c:130) [ 721.016779][ C0] panic (kernel/panic.c:258 kernel/panic.c:375) [ 721.016779][ C0] watchdog_timer_fn (kernel/watchdog.c:740) [ 721.016779][ C0] ? schedule_work (drivers/usb/core/hub.c:919) [ 721.016779][ C0] __hrtimer_run_queues+0x12f/0x1cf [ 721.016779][ C0] hrtimer_run_queues (kernel/time/hrtimer.c:2023) [ 721.016779][ C0] update_process_times (kernel/time/timer.c:2458 kernel/time/timer.c:2514) [ 721.016779][ C0] tick_periodic (kernel/time/tick-common.c:103) [ 721.016779][ C0] tick_handle_periodic (kernel/time/tick-common.c:144) [ 721.016779][ C0] ? vmware_sched_clock (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] __sysvec_apic_timer_interrupt (arch/x86/include/asm/trace/irq_vectors.h:41 arch/x86/include/asm/trace/irq_vectors.h:41 arch/x86/kernel/apic/apic.c:1056) [ 721.016779][ C0] sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049 arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] handle_exception (arch/x86/entry/entry_32.S:1055) [ 721.016779][ C0] EIP: timekeeping_notify (kernel/time/timekeeping.c:1522) [ 721.016779][ C0] Code: 5f e9 ff ff 8d 45 e8 e8 57 d4 ff ff 85 ff 74 16 8b 57 5c 85 d2 74 04 89 f8 ff d2 8b 87 88 00 00 00 e8 d5 3e ff ff 85 f6 75 9b <e8> 7f b9 00 00 31 c0 39 1d a4 70 14 84 0f 95 c0 f7 d8 8b 55 f0 2b All code ======== 0: 5f pop %rdi 1: e9 ff ff 8d 45 jmp 0x458e0005 6: e8 e8 57 d4 ff call 0xffffffffffd457f3 b: ff 85 ff 74 16 8b incl -0x74e98b01(%rbp) 11: 57 push %rdi 12: 5c pop %rsp 13: 85 d2 test %edx,%edx 15: 74 04 je 0x1b 17: 89 f8 mov %edi,%eax 19: ff d2 call *%rdx 1b: 8b 87 88 00 00 00 mov 0x88(%rdi),%eax 21: e8 d5 3e ff ff call 0xffffffffffff3efb 26: 85 f6 test %esi,%esi 28: 75 9b jne 0xffffffffffffffc5 2a:* e8 7f b9 00 00 call 0xb9ae <-- trapping instruction 2f: 31 c0 xor %eax,%eax 31: 39 1d a4 70 14 84 cmp %ebx,-0x7beb8f5c(%rip) # 0xffffffff841470db 37: 0f 95 c0 setne %al 3a: f7 d8 neg %eax 3c: 8b 55 f0 mov -0x10(%rbp),%edx 3f: 2b .byte 0x2b
Code starting with the faulting instruction
===========================================
0: e8 7f b9 00 00 call 0xb984 5: 31 c0 xor %eax,%eax 7: 39 1d a4 70 14 84 cmp %ebx,-0x7beb8f5c(%rip) # 0xffffffff841470b1 d: 0f 95 c0 setne %al 10: f7 d8 neg %eax 12: 8b 55 f0 mov -0x10(%rbp),%edx 15: 2b .byte 0x2b
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20250421/[email protected]
[1]
pkg/linux/i386-randconfig-r071-20250410/gcc-12/f388f60ca9041a95c9b3f157d316ed7c8f297e44/.config 2025-04-15 15:41:17.009901645 +0800 renamed from pkg/linux/i386-randconfig-r071-20250410/gcc-12/fc2d5cbe541032e74a66599ba843803cebbfed0e/.config 2025-04-15 15:41:11.316836213 +0800 +13 -22
@@ -321,7 +321,7 @@ CONFIG_ARCH_CPUIDLE_HALTPOLL=y # CONFIG_PVH is not set # CONFIG_PARAVIRT_TIME_ACCOUNTING is not set CONFIG_PARAVIRT_CLOCK=y -# CONFIG_M486SX is not set +CONFIG_M486SX=y # CONFIG_M486 is not set # CONFIG_M586 is not set # CONFIG_M586TSC is not set
@@ -333,7 +333,6 @@ CONFIG_PARAVIRT_CLOCK=y # CONFIG_MPENTIUM4 is not set # CONFIG_MK6 is not set # CONFIG_MK7 is not set -CONFIG_MK8=y # CONFIG_MCRUSOE is not set # CONFIG_MEFFICEON is not set # CONFIG_MWINCHIPC6 is not set
@@ -344,26 +343,24 @@ CONFIG_MK8=y # CONFIG_MCYRIXIII is not set # CONFIG_MVIAC3_2 is not set # CONFIG_MVIAC7 is not set -# CONFIG_MCORE2 is not set # CONFIG_MATOM is not set CONFIG_X86_GENERIC=y CONFIG_X86_INTERNODE_CACHE_SHIFT=6 CONFIG_X86_L1_CACHE_SHIFT=6 +CONFIG_X86_F00F_BUG=y +CONFIG_X86_INVD_BUG=y +CONFIG_X86_ALIGNMENT_16=y CONFIG_X86_INTEL_USERCOPY=y -CONFIG_X86_USE_PPRO_CHECKSUM=y -CONFIG_X86_TSC=y -CONFIG_X86_HAVE_PAE=y -CONFIG_X86_CMPXCHG64=y -CONFIG_X86_CMOV=y -CONFIG_X86_MINIMUM_CPU_FAMILY=6 -CONFIG_X86_DEBUGCTLMSR=y +CONFIG_X86_MINIMUM_CPU_FAMILY=4 CONFIG_IA32_FEAT_CTL=y CONFIG_X86_VMX_FEATURE_NAMES=y CONFIG_CPU_SUP_INTEL=y +CONFIG_CPU_SUP_CYRIX_32=y CONFIG_CPU_SUP_AMD=y CONFIG_CPU_SUP_HYGON=y CONFIG_CPU_SUP_CENTAUR=y CONFIG_CPU_SUP_TRANSMETA_32=y +CONFIG_CPU_SUP_UMC_32=y CONFIG_CPU_SUP_ZHAOXIN=y CONFIG_CPU_SUP_VORTEX_32=y CONFIG_HPET_TIMER=y
@@ -410,7 +407,6 @@ CONFIG_X86_MSR=y # CONFIG_X86_CPUID is not set # CONFIG_NOHIGHMEM is not set CONFIG_HIGHMEM4G=y -# CONFIG_HIGHMEM64G is not set # CONFIG_VMSPLIT_3G is not set # CONFIG_VMSPLIT_3G_OPT is not set CONFIG_VMSPLIT_2G=y
@@ -418,7 +414,6 @@ CONFIG_VMSPLIT_2G=y # CONFIG_VMSPLIT_1G is not set CONFIG_PAGE_OFFSET=0x80000000 CONFIG_HIGHMEM=y -# CONFIG_X86_PAE is not set # CONFIG_X86_CPA_STATISTICS is not set CONFIG_ARCH_FLATMEM_ENABLE=y CONFIG_ARCH_SPARSEMEM_ENABLE=y
@@ -427,6 +422,7 @@ CONFIG_ILLEGAL_POINTER_VALUE=0 # CONFIG_HIGHPTE is not set CONFIG_X86_CHECK_BIOS_CORRUPTION=y CONFIG_X86_BOOTPARAM_MEMORY_CORRUPTION_CHECK=y +# CONFIG_MATH_EMULATION is not set CONFIG_MTRR=y CONFIG_MTRR_SANITIZER=y CONFIG_MTRR_SANITIZER_ENABLE_DEFAULT=0
@@ -472,8 +468,8 @@ CONFIG_USE_X86_SEG_SUPPORT=y CONFIG_CC_HAS_SLS=y CONFIG_CC_HAS_RETURN_THUNK=y CONFIG_CC_HAS_ENTRY_PADDING=y -CONFIG_FUNCTION_PADDING_CFI=0 -CONFIG_FUNCTION_PADDING_BYTES=4 +CONFIG_FUNCTION_PADDING_CFI=11 +CONFIG_FUNCTION_PADDING_BYTES=16 CONFIG_CPU_MITIGATIONS=y # CONFIG_MITIGATION_RETPOLINE is not set # CONFIG_MITIGATION_GDS is not set
@@ -741,7 +737,8 @@ CONFIG_ARCH_HAS_GCOV_PROFILE_ALL=y CONFIG_HAVE_GCC_PLUGINS=y # CONFIG_GCC_PLUGINS is not set CONFIG_FUNCTION_ALIGNMENT_4B=y -CONFIG_FUNCTION_ALIGNMENT=4 +CONFIG_FUNCTION_ALIGNMENT_16B=y +CONFIG_FUNCTION_ALIGNMENT=16 # end of General architecture-dependent options CONFIG_RT_MUTEXES=y
@@ -1114,7 +1111,6 @@ CONFIG_NFC_SHDLC=y # # Near Field Communication (NFC) devices # -# CONFIG_NFC_MEI_PHY is not set # CONFIG_NFC_SIM is not set # CONFIG_NFC_PORT100 is not set # CONFIG_NFC_PN544_I2C is not set
@@ -1607,9 +1603,7 @@ CONFIG_EEPROM_IDT_89HPESX=y # CONFIG_CB710_CORE is not set CONFIG_SENSORS_LIS3_I2C=y CONFIG_ALTERA_STAPL=y -CONFIG_INTEL_MEI=y -CONFIG_INTEL_MEI_ME=y -# CONFIG_INTEL_MEI_TXE is not set +# CONFIG_INTEL_MEI is not set # CONFIG_VMWARE_VMCI is not set CONFIG_ECHO=y # CONFIG_MISC_ALCOR_PCI is not set
@@ -3412,7 +3406,6 @@ CONFIG_TQMX86_WDT=y CONFIG_W83977F_WDT=y CONFIG_MACHZ_WDT=y CONFIG_SBC_EPX_C3_WATCHDOG=y -# CONFIG_INTEL_MEI_WDT is not set CONFIG_NI903X_WDT=y # CONFIG_NIC7018_WDT is not set # CONFIG_MEN_A21_WDT is not set
@@ -5752,7 +5745,6 @@ CONFIG_GENERIC_NET_UTILS=y # CONFIG_PRIME_NUMBERS is not set CONFIG_RATIONAL=y CONFIG_GENERIC_IOMAP=y -CONFIG_ARCH_USE_CMPXCHG_LOCKREF=y CONFIG_ARCH_HAS_FAST_MULTIPLIER=y CONFIG_ARCH_USE_SYM_ANNOTATIONS=y
@@ -6186,7 +6178,6 @@ CONFIG_SAMPLE_VFIO_MDEV_MDPY=y CONFIG_SAMPLE_VFIO_MDEV_MBOCHS=y CONFIG_SAMPLE_ANDROID_BINDERFS=y CONFIG_SAMPLE_VFS=y -# CONFIG_SAMPLE_INTEL_MEI is not set # CONFIG_SAMPLE_TPS6594_PFSM is not set CONFIG_SAMPLE_WATCHDOG=y CONFIG_SAMPLE_WATCH_QUEUE=y
Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
Quoted 6 lines · kernel test robot
Hello,
by this commit, we notice big config diff [1]
then in this rcutorture tests, parent runs quite clean,
f388f60ca9shows various random issues.
Thanks for the report!
From my initial reading, my patch most likely caught a preexisting bug, but my patch itself is correct. It's worth investigating regardless, at the minimum we should perhaps prevent an invalid configuration from building or from booting.
Quoted 1 line
config: i386-randconfig-r071-20250410
Generally, I would not expect 'randconfig' kernels to pass all tests, and what happened here is that removing the CONFIG_MK8 option made it pick some completely different CPU
Quoted 2 lines
compiler: gcc-12
test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16G
The most relevant options here are
-# CONFIG_M486SX is not set
+CONFIG_M486SX=y
# CONFIG_SMP is not set
CONFIG_X86_GENERIC=y
In theory, setting X86_GENERIC should make a kernel built for an older CPU work on any newer one. In practice, I'm not surprised that this fails: While AMD K8 is ten years older than Intel Sandy Bridge, they are architecturally still very similar. The i486SX is another decade older, but its design is as far removed from both K8 and Sandy Bridge as it gets.
It would be nice to not have to support 486sx any more. We have discussed removing support for older CPUs without TSC, FPU and CX8 in the past, but so far always kept them around.
Quoted 11 lines
[ 721.016779][ C0] hardirqs last disabled at (159506): sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] softirqs last enabled at (159174): handle_softirqs (kernel/softirq.c:408 kernel/softirq.c:589) [ 721.016779][ C0] softirqs last disabled at (159159): __do_softirq (kernel/softirq.c:596) [ 721.016779][ C0] CPU: 0 UID: 0 PID: 1 Comm: swapper Not tainted 6.14.0-rc3-00037-gf388f60ca904 #1 [ 721.016779][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 [ 721.016779][ C0] EIP: timekeeping_notify
Quoted 1 line
(
kernel/time/timekeeping.c:1522)
Timekeeping code could be related, I see that CONFIG_X86_TSC is disabled for i486SX configurations, so even if a TSC is present in the emulated machine, it is not being used to measure time accurately.
Quoted 1 line
-CONFIG_X86_CMPXCHG64=y
This could be another issue, if there is code that relies on the cx8/cmpxchg8b feature to be used. Since this is a non-SMP kernel, this is less likely to be the cause of the problem.
Can you try what happens when you enable the two options, either by changing CONFIG_M486SX to CONFIG_M586TSC, or with a patch like the one below? Note that CONFIG_X86_CMPXCHG64 recently got renamed to CONFIG_X86_CX8, but they are the exact same thing.
arch/x86/Kconfig.cpu +0 -2
@@ -317,7 +317,6 @@ config X86_USE_PPRO_CHECKSUM config X86_TSC def_bool y - depends on (MWINCHIP3D || MCRUSOE || MEFFICEON || MCYRIXIII || MK7 || MK6 || MPENTIUM4 || MPENTIUMM || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || MVIAC3_2 || MVIAC7 || MGEODEGX1 || MGEODE_LX || MATOM) || X86_64 config X86_HAVE_PAE def_bool y
@@ -325,7 +324,6 @@ config X86_HAVE_PAE config X86_CX8 def_bool y - depends on X86_HAVE_PAE || M586TSC || M586MMX || MK6 || MK7 || MGEODEGX1 || MGEODE_LX # this should be set for all -march=.. options where the compiler # generates cmov.
Arnd
hi, Arnd,
Quoted 61 lines · Arnd Bergmann
Quoted 6 lines · kernel test robot
Hello,
by this commit, we notice big config diff [1]
then in this rcutorture tests, parent runs quite clean,
f388f60ca9shows various random issues.Thanks for the report!
From my initial reading, my patch most likely caught a preexisting bug, but my patch itself is correct. It's worth investigating regardless, at the minimum we should perhaps prevent an invalid configuration from building or from booting.
Quoted 1 line
config: i386-randconfig-r071-20250410
Generally, I would not expect 'randconfig' kernels to pass all tests, and what happened here is that removing the
CONFIG_MK8option made it pick some completely different CPUQuoted 2 lines
compiler: gcc-12
test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16GThe most relevant options here are
-#
CONFIG_M486SXis not set
+CONFIG_M486SX=y
#CONFIG_SMPis not set
CONFIG_X86_GENERIC=yIn theory, setting
X86_GENERICshould make a kernel built for an older CPU work on any newer one. In practice, I'm not surprised that this fails: While AMD K8 is ten years older than Intel Sandy Bridge, they are architecturally still very similar. The i486SX is another decade older, but its design is as far removed from both K8 and Sandy Bridge as it gets.It would be nice to not have to support 486sx any more. We have discussed removing support for older CPUs without TSC, FPU and CX8 in the past, but so far always kept them around.
Quoted 11 lines
[ 721.016779][ C0] hardirqs last disabled at (159506): sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] softirqs last enabled at (159174): handle_softirqs (kernel/softirq.c:408 kernel/softirq.c:589) [ 721.016779][ C0] softirqs last disabled at (159159): __do_softirq (kernel/softirq.c:596) [ 721.016779][ C0] CPU: 0 UID: 0 PID: 1 Comm: swapper Not tainted 6.14.0-rc3-00037-gf388f60ca904 #1 [ 721.016779][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 [ 721.016779][ C0] EIP: timekeeping_notifyQuoted 1 line
(
kernel/time/timekeeping.c:1522)Timekeeping code could be related, I see that
CONFIG_X86_TSCis disabled for i486SX configurations, so even if a TSC is present in the emulated machine, it is not being used to measure time accurately.
Quoted 5 lines
Quoted 1 line
-CONFIG_X86_CMPXCHG64=y
This could be another issue, if there is code that relies on the cx8/cmpxchg8b feature to be used. Since this is a non-SMP kernel, this is less likely to be the cause of the problem.
thanks a lot for all these details!
Quoted 5 lines
Can you try what happens when you enable the two options, either by changing
CONFIG_M486SXtoCONFIG_M586TSC, or with a patch like the one below? Note thatCONFIG_X86_CMPXCHG64recently got renamed toCONFIG_X86_CX8, but they are the exact same thing.
I applied your patch directly upon f388f60ca9 (change for X86_CMPXCHG64 instead of X86_CX8 as you metnioned), commit id is c1f7ef63239411313163a7b1bff654236f48351c
after building, the config has below diff to f388f60ca9
.config 2025-04-23 09:36:43.718421931 +0800 renamed from .config 2025-04-15 15:41:17.009901645 +0800 +4 -1
@@ -351,7 +351,9 @@ CONFIG_X86_F00F_BUG=y CONFIG_X86_INVD_BUG=y CONFIG_X86_ALIGNMENT_16=y CONFIG_X86_INTEL_USERCOPY=y -CONFIG_X86_MINIMUM_CPU_FAMILY=4 +CONFIG_X86_TSC=y +CONFIG_X86_CMPXCHG64=y +CONFIG_X86_MINIMUM_CPU_FAMILY=5 CONFIG_IA32_FEAT_CTL=y CONFIG_X86_VMX_FEATURE_NAMES=y CONFIG_CPU_SUP_INTEL=y
@@ -5745,6 +5747,7 @@ CONFIG_GENERIC_NET_UTILS=y # CONFIG_PRIME_NUMBERS is not set CONFIG_RATIONAL=y CONFIG_GENERIC_IOMAP=y +CONFIG_ARCH_USE_CMPXCHG_LOCKREF=y CONFIG_ARCH_HAS_FAST_MULTIPLIER=y CONFIG_ARCH_USE_SYM_ANNOTATIONS=y
by running same tests, now it backs to the clean status like fc2d5cbe541032e7 (parent of f388f60ca9)
(the statistics data for fc2d5cbe541032e7 and f388f60ca9 has some difference to the data we shared last time due to some auto cleanup logic in our service which removes some results which are suspiciously caused by our env problem)
fc2d5cbe541032e7 f388f60ca9041a95c9b3f157d31 c1f7ef63239411313163a7b1bff ---------------- --------------------------- ---------------------------
fail:runs %reproduction fail:runs %reproduction fail:runs
| | | | |
:496 29% 145:494 0% :500 last_state.booting
:496 7% 35:494 0% :500 dmesg.BUG:kernel_hang_in_boot_stage
:496 9% 45:494 0% :500 dmesg.BUG:soft_lockup-CPU##stuck_for#s![swapper:#]
:496 0% 1:494 0% :500 dmesg.EIP:__timer_delete_sync
:496 1% 5:494 0% :500 dmesg.EIP:_raw_spin_unlock_irq
:496 0% 2:494 0% :500 dmesg.EIP:_raw_spin_unlock_irqrestore
:496 0% 1:494 0% :500 dmesg.EIP:console_emit_next_record
:496 0% 1:494 0% :500 dmesg.EIP:handle_softirqs
:496 1% 3:494 0% :500 dmesg.EIP:lock_acquire
:496 0% 2:494 0% :500 dmesg.EIP:lock_release
:496 0% 1:494 0% :500 dmesg.EIP:queue_delayed_work_on
:496 9% 45:494 0% :500 dmesg.EIP:timekeeping_notify
:496 3% 14:494 0% :500 dmesg.INFO:rcu_preempt_detected_stalls_on_CPUs/tasks
:496 6% 32:494 0% :500 dmesg.INFO:task_blocked_for_more_than#seconds
:496 9% 45:494 0% :500 dmesg.Kernel_panic-not_syncing:softlockup:hung_tasks
Quoted 23 lines
arch/x86/Kconfig.cpu +0
-2@@ -317,7 +317,6 @@ config X86_USE_PPRO_CHECKSUM config X86_TSC def_bool y - depends on (MWINCHIP3D || MCRUSOE || MEFFICEON || MCYRIXIII || MK7 || MK6 || MPENTIUM4 || MPENTIUMM || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || MVIAC3_2 || MVIAC7 || MGEODEGX1 || MGEODE_LX || MATOM) || X86_64 config X86_HAVE_PAE def_bool y@@ -325,7 +324,6 @@ config X86_HAVE_PAE config X86_CX8 def_bool y - depends on X86_HAVE_PAE || M586TSC || M586MMX || MK6 || MK7 || MGEODEGX1 || MGEODE_LX # this should be set for all -march=.. options where the compiler # generates cmov.Arnd
Quoted 1 line · Oliver Sang
On Tue, Apr 22, 2025 at 12:16:33PM +0200, Arnd Bergmann wrote:
Cc: x86 and timekeeping maintainers, see
https://lore.kernel.org/lkml/[email protected]/
for the thread so far.
Quoted 36 lines
Quoted 18 lines
Quoted 11 lines
[ 721.016779][ C0] hardirqs last disabled at (159506): sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] softirqs last enabled at (159174): handle_softirqs (kernel/softirq.c:408 kernel/softirq.c:589) [ 721.016779][ C0] softirqs last disabled at (159159): __do_softirq (kernel/softirq.c:596) [ 721.016779][ C0] CPU: 0 UID: 0 PID: 1 Comm: swapper Not tainted 6.14.0-rc3-00037-gf388f60ca904 #1 [ 721.016779][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 [ 721.016779][ C0] EIP: timekeeping_notifyQuoted 1 line
(
kernel/time/timekeeping.c:1522)Timekeeping code could be related, I see that
CONFIG_X86_TSCis disabled for i486SX configurations, so even if a TSC is present in the emulated machine, it is not being used to measure time accurately.Quoted 5 lines
Quoted 1 line
-CONFIG_X86_CMPXCHG64=y
This could be another issue, if there is code that relies on the cx8/cmpxchg8b feature to be used. Since this is a non-SMP kernel, this is less likely to be the cause of the problem.
thanks a lot for all these details!
Quoted 5 lines
Can you try what happens when you enable the two options, either by changing
CONFIG_M486SXtoCONFIG_M586TSC, or with a patch like the one below? Note thatCONFIG_X86_CMPXCHG64recently got renamed toCONFIG_X86_CX8, but they are the exact same thing.I applied your patch directly upon
f388f60ca9(change forX86_CMPXCHG64instead ofX86_CX8as you metnioned), commit id isc1f7ef63239411313163a7b1bff654236f48351c
...
Quoted 2 lines
by running same tests, now it backs to the clean status like
fc2d5cbe541032e7(parent off388f60ca9)
Thanks for confirming. So a 486-targeted kernel still passes your tests on modern hardware if we force TSC and CX8 to be enabled, but the boot fails if the options are turned off in Kconfig (though available in emulated hardware).
To be completely sure, you could re-run the same test with just one of these enabled, but I'm rather sure that the TSC is the root cause. I tried reproducing the problem locally with your .config on a qemu/tcg emulation running on an arm64 host, but this seems to run fine, including the rcutorture tests.
Comparing my results with your log file, I see that your crash happens while changing the clocksource:
Your dmesg:
[ 92.548514][ T1] hpet0: 3 comparators, 64-bit 100.000000 MHz counter [ 721.016745][ C0] watchdog: BUG: soft lockup - CPU#0 stuck for 626s! [swapper:1]
My dmesg:
[ 1.154511][ T1] hpet0: 3 comparators, 64-bit 100.000000 MHz counter [ 1.157896][ T1] clocksource: Switched to clocksource tsc-early
There are also clearly some differences between TCG and KVM in the handling of TSC, e.g. I get this warning from qemu itself for the SandyBridge CPU:
qemu-system-i386: warning: TCG doesn't support requested feature: CPUID.01H:ECX.tsc-deadline [bit 24]
I tried a few other variations, including KVM on an x86 laptop (using kvmclock or tsc-early clocksource), but none of them failed the way yours did.
Arnd
Quoted 5 lines · Arnd Bergmann
Thanks for confirming. So a 486-targeted kernel still passes your tests on modern hardware if we force TSC and CX8 to be enabled, but the boot fails if the options are turned off in
Kconfig(though available in emulated hardware).
I wouldn't expect CX8 to really matter - it causes us to generate extra code to pick one over the other, but on modern hardware we'll still always then dynamically pick the cmpxchg8b instruction.
Could it trigger bugs in our alternatives, or some miscompilation due to the extra complexity? Sure. But it does sound unlikely.
Quoted 3 lines
To be completely sure, you could re-run the same test with just one of these enabled, but I'm rather sure that the TSC is the root cause.
Agreed.
Particularly when the lockup is then in timekeeping_notify() during the initial initcalls -> clocksource_select(), I'm pretty sure this is purely about TSC.
That said, maybe the problem is in the watchdog logic, because clocksource_done_booting() is what starts the watchdog thread .
So it might be the watchdog code itself that then gets confused (because of some "don't use tsc" case that never gets any testing in real life) and triggers immediately - and then points the finger at the clocksource code only because that's what is still running.
Because CONFIG_X86_TSC does cause some oddities: we end up still *using* the TSC for many things if the hardware supports it (which modern hardware obviously does), but then other things get disabled entirely.
For example, this:
/*
* Boot-time check whether the TSCs are synchronized across
* all CPUs/cores:
*/
#ifdef CONFIG_X86_TSC
extern bool tsc_store_and_check_tsc_adjust(bool bootcpu); extern void tsc_verify_tsc_adjust(bool resume);
extern void check_tsc_sync_target(void);
#else
static inline bool tsc_store_and_check_tsc_adjust(bool bootcpu) {
return false; }
static inline void tsc_verify_tsc_adjust(bool resume) { } static inline void check_tsc_sync_target(void) { } #endif
So that tsc_store_and_check_tsc_adjust() thing etc never gets run, even though we actually *do* use TSC for get_cycles() and friends, because *that* code checks the runtime status too:
Now, none of that should matter - because all *those* things are about details that simply aren't relevant for any of this case - but maybe there is some other situation that has similar "I'm actually using the TSC through get_cycles(), but I didn't do some setup because X86_TSC wasn't on.."
I really get the feeling that it's time to leave i486 support behind. There's zero real reason for anybody to waste one second of development effort on this kind of issue.
Linus
cover x86: Remove support for TSC-less and CX8-less CPUs
Quoted 3 lines · Linus Torvalds
I really get the feeling that it's time to leave i486 support behind. There's zero real reason for anybody to waste one second of development effort on this kind of issue.
Fully agreed!
And to turn this idea into code, here's a very raw RFC series that starts removing non-TSC 586 and 486 code and related support code from the x86 architecture, with the goal to make TSC and CX8 (CMPXCHG8B) support unconditionally available:
git://git.kernel.org/pub/scm/linux/kernel/git/mingo/tip.git WIP.x86/cpu
The full diffstat is nice, primarily due to the removal of the math-emu library:
83 files changed, 30 insertions(+), 14683 deletions(-)
But even without the math-emu/ removal and the drivers/ pruning it's a substantial simplification:
20 files changed, 29 insertions(+), 629 deletions(-)
The patches most relevant to this discussion should be:
x86/cpu: Remove M486/M486SX/ELAN support
...
x86/cpu: Remove TSC-less CONFIG_M586 support x86/cpu: Make CONFIG_X86_TSC unconditional x86/cpu: Make CONFIG_X86_CX8 unconditional x86/percpu: Remove !CONFIG_X86_CX8 methods x86/atomics: Remove !CONFIG_X86_CX8 methods
If there's no big objections about the scope of removal I'll finish it by removing the non-TSC complications as well, and send out the series to lkml for further review.
( Note that some of the patches in there are still WIP, as the branch
name suggests. )
Thanks,
Ingo
================>
Ingo Molnar (17):
x86/cpu: Remove M486/M486SX/ELAN support x86/cpu: Remove the CONFIG_X86_INVD_BUG quirk x86/cpu, cpufreq: Remove AMD ELAN support x86/fpu: Remove MATH_EMULATION and related glue code x86/fpu: Remove the 'no387' boot option
x86/fpu: Remove the math-emu/ FPU emulation library x86/platform: Remove CONFIG_X86_RDC321X support arch/x86, gpio: Remove GPIO_RDC321X support arch/x86, watchdog: Remove the RDC321X_WDT watchdog driver arch/x86, mfd: Remove MFD_RDC321X support x86/reboot: Remove the RDC321X reboot quirk x86/cpu: Remove CPU_SUP_UMC_32 support
x86/cpu: Remove TSC-less CONFIG_M586 support x86/cpu: Make CONFIG_X86_TSC unconditional x86/cpu: Make CONFIG_X86_CX8 unconditional x86/percpu: Remove !X86_CX8 methods
x86/atomics: Remove !CONFIG_X86_CX8 methods
Documentation/admin-guide/kernel-parameters.txt | 4 - arch/x86/Kconfig | 71 +- arch/x86/Kconfig.cpu | 73 +- arch/x86/Kconfig.cpufeatures | 2 - arch/x86/Makefile | 1 - arch/x86/Makefile_32.cpu | 6 - arch/x86/include/asm/asm-prototypes.h | 4 - arch/x86/include/asm/atomic64_32.h | 17 +- arch/x86/include/asm/cmpxchg_32.h | 86 +- arch/x86/include/asm/fpu/api.h | 6 - arch/x86/include/asm/percpu.h | 6 +- arch/x86/include/asm/vermagic.h | 8 - arch/x86/kernel/cpu/common.c | 7 - arch/x86/kernel/cpu/umc.c | 26 - arch/x86/kernel/fpu/core.c | 5 - arch/x86/kernel/fpu/init.c | 9 +- arch/x86/kernel/reboot_fixups_32.c | 14 - arch/x86/kernel/traps.c | 21 - arch/x86/lib/Makefile | 4 - arch/x86/lib/atomic64_386_32.S | 195 --- arch/x86/lib/cmpxchg8b_emu.S | 97 -- arch/x86/math-emu/Makefile | 30 - arch/x86/math-emu/README | 427 ------ arch/x86/math-emu/control_w.h | 46 - arch/x86/math-emu/div_Xsig.S | 367 ----- arch/x86/math-emu/div_small.S | 48 - arch/x86/math-emu/errors.c | 686 ---------- arch/x86/math-emu/exception.h | 51 - arch/x86/math-emu/fpu_arith.c | 153 --- arch/x86/math-emu/fpu_asm.h | 32 - arch/x86/math-emu/fpu_aux.c | 267 ---- arch/x86/math-emu/fpu_emu.h | 218 --- arch/x86/math-emu/fpu_entry.c | 718 ---------- arch/x86/math-emu/fpu_etc.c | 136 -- arch/x86/math-emu/fpu_proto.h | 157 --- arch/x86/math-emu/fpu_system.h | 130 -- arch/x86/math-emu/fpu_tags.c | 116 -- arch/x86/math-emu/fpu_trig.c | 1649 ----------------------- arch/x86/math-emu/get_address.c | 401 ------ arch/x86/math-emu/load_store.c | 322 ----- arch/x86/math-emu/mul_Xsig.S | 179 --- arch/x86/math-emu/poly.h | 115 -- arch/x86/math-emu/poly_2xm1.c | 146 -- arch/x86/math-emu/poly_atan.c | 209 --- arch/x86/math-emu/poly_l2.c | 245 ---- arch/x86/math-emu/poly_sin.c | 379 ------ arch/x86/math-emu/poly_tan.c | 213 --- arch/x86/math-emu/polynom_Xsig.S | 137 -- arch/x86/math-emu/reg_add_sub.c | 334 ----- arch/x86/math-emu/reg_compare.c | 479 ------- arch/x86/math-emu/reg_constant.c | 123 -- arch/x86/math-emu/reg_constant.h | 26 - arch/x86/math-emu/reg_convert.c | 47 - arch/x86/math-emu/reg_divide.c | 183 --- arch/x86/math-emu/reg_ld_str.c | 1220 ----------------- arch/x86/math-emu/reg_mul.c | 116 -- arch/x86/math-emu/reg_norm.S | 150 --- arch/x86/math-emu/reg_round.S | 711 ---------- arch/x86/math-emu/reg_u_add.S | 169 --- arch/x86/math-emu/reg_u_div.S | 474 ------- arch/x86/math-emu/reg_u_mul.S | 150 --- arch/x86/math-emu/reg_u_sub.S | 274 ---- arch/x86/math-emu/round_Xsig.S | 142 -- arch/x86/math-emu/shr_Xsig.S | 89 -- arch/x86/math-emu/status_w.h | 68 - arch/x86/math-emu/version.h | 12 - arch/x86/math-emu/wm_shrx.S | 207 --- arch/x86/math-emu/wm_sqrt.S | 472 ------- drivers/cpufreq/Kconfig.x86 | 26 - drivers/cpufreq/Makefile | 2 - drivers/cpufreq/elanfreq.c | 227 ---- drivers/cpufreq/sc520_freq.c | 137 -- drivers/gpio/Kconfig | 8 - drivers/gpio/Makefile | 1 - drivers/gpio/gpio-rdc321x.c | 197 --- drivers/mfd/Kconfig | 9 - drivers/mfd/Makefile | 1 - drivers/mfd/rdc321x-southbridge.c | 96 -- drivers/watchdog/Kconfig | 11 - drivers/watchdog/Makefile | 1 - drivers/watchdog/rdc321x_wdt.c | 281 ---- include/linux/mfd/rdc321x.h | 27 - lib/atomic64_test.c | 4 +- 83 files changed, 30 insertions(+), 14683 deletions(-) delete mode 100644 arch/x86/kernel/cpu/umc.c delete mode 100644 arch/x86/lib/atomic64_386_32.S delete mode 100644 arch/x86/lib/cmpxchg8b_emu.S delete mode 100644 arch/x86/math-emu/Makefile delete mode 100644 arch/x86/math-emu/README delete mode 100644 arch/x86/math-emu/control_w.h delete mode 100644 arch/x86/math-emu/div_Xsig.S delete mode 100644 arch/x86/math-emu/div_small.S delete mode 100644 arch/x86/math-emu/errors.c delete mode 100644 arch/x86/math-emu/exception.h delete mode 100644 arch/x86/math-emu/fpu_arith.c delete mode 100644 arch/x86/math-emu/fpu_asm.h delete mode 100644 arch/x86/math-emu/fpu_aux.c delete mode 100644 arch/x86/math-emu/fpu_emu.h delete mode 100644 arch/x86/math-emu/fpu_entry.c delete mode 100644 arch/x86/math-emu/fpu_etc.c delete mode 100644 arch/x86/math-emu/fpu_proto.h delete mode 100644 arch/x86/math-emu/fpu_system.h delete mode 100644 arch/x86/math-emu/fpu_tags.c delete mode 100644 arch/x86/math-emu/fpu_trig.c delete mode 100644 arch/x86/math-emu/get_address.c delete mode 100644 arch/x86/math-emu/load_store.c delete mode 100644 arch/x86/math-emu/mul_Xsig.S delete mode 100644 arch/x86/math-emu/poly.h delete mode 100644 arch/x86/math-emu/poly_2xm1.c delete mode 100644 arch/x86/math-emu/poly_atan.c delete mode 100644 arch/x86/math-emu/poly_l2.c delete mode 100644 arch/x86/math-emu/poly_sin.c delete mode 100644 arch/x86/math-emu/poly_tan.c delete mode 100644 arch/x86/math-emu/polynom_Xsig.S delete mode 100644 arch/x86/math-emu/reg_add_sub.c delete mode 100644 arch/x86/math-emu/reg_compare.c delete mode 100644 arch/x86/math-emu/reg_constant.c delete mode 100644 arch/x86/math-emu/reg_constant.h delete mode 100644 arch/x86/math-emu/reg_convert.c delete mode 100644 arch/x86/math-emu/reg_divide.c delete mode 100644 arch/x86/math-emu/reg_ld_str.c delete mode 100644 arch/x86/math-emu/reg_mul.c delete mode 100644 arch/x86/math-emu/reg_norm.S delete mode 100644 arch/x86/math-emu/reg_round.S delete mode 100644 arch/x86/math-emu/reg_u_add.S delete mode 100644 arch/x86/math-emu/reg_u_div.S delete mode 100644 arch/x86/math-emu/reg_u_mul.S delete mode 100644 arch/x86/math-emu/reg_u_sub.S delete mode 100644 arch/x86/math-emu/round_Xsig.S delete mode 100644 arch/x86/math-emu/shr_Xsig.S delete mode 100644 arch/x86/math-emu/status_w.h delete mode 100644 arch/x86/math-emu/version.h delete mode 100644 arch/x86/math-emu/wm_shrx.S delete mode 100644 arch/x86/math-emu/wm_sqrt.S delete mode 100644 drivers/cpufreq/elanfreq.c delete mode 100644 drivers/cpufreq/sc520_freq.c delete mode 100644 drivers/gpio/gpio-rdc321x.c delete mode 100644 drivers/mfd/rdc321x-southbridge.c delete mode 100644 drivers/watchdog/rdc321x_wdt.c delete mode 100644 include/linux/mfd/rdc321x.h
Re: [RFC PATCH 0/15] x86: Remove support for TSC-less and CX8-less CPUs
Quoted 9 lines · Ingo Molnar
The patches most relevant to this discussion should be:
x86/cpu: Remove M486/M486SX/ELAN support
...
x86/cpu: Remove TSC-lessCONFIG_M586support x86/cpu: MakeCONFIG_X86_TSCunconditional x86/cpu: MakeCONFIG_X86_CX8unconditional x86/percpu: Remove !CONFIG_X86_CX8 methods x86/atomics: Remove !CONFIG_X86_CX8 methods
also:
Quoted 1 line
x86/cpu: Remove TSC-less
CONFIG_M586support
Thanks,
Ingo
Re: [RFC PATCH 0/15] x86: Remove support for TSC-less and CX8-less CPUs
In reply to Ingo MolnarQuoted 6 lines · Ingo Molnar
Quoted 3 lines · Linus Torvalds
I really get the feeling that it's time to leave i486 support behind. There's zero real reason for anybody to waste one second of development effort on this kind of issue.
Quoted 6 lines
Fully agreed!
And to turn this idea into code, here's a very raw RFC series that starts removing non-TSC 586 and 486 code and related support code from the x86 architecture, with the goal to make TSC and CX8 (CMPXCHG8B) support unconditionally available:
Yes, makes sense. I had considered doing something like this in my cleanup for large machines, but decided to keep stop there because I know that there are users that love their museum pieces.
For embedded systems, I'm quite sure that the AMD Élan, SIS 55x, RDC321x and Vortex86SX have ended their useful life. You may still be able to buy Vortex86SX machines and they are probably still running somewhere, but the entire point of those machines is to run old software. DM&P provides patches for linux-2.6.29 and Windows CE 6.0 for the SX chips.
The desktop Socket 7 clones without CX8/TSC all got discontinued over 25 years ago, and they were rare even then.
Quoted 5 lines
x86/platform: Remove
CONFIG_X86_RDC321Xsupportarch/x86, gpio: RemoveGPIO_RDC321Xsupportarch/x86, watchdog: Remove theRDC321X_WDTwatchdog driverarch/x86, mfd: RemoveMFD_RDC321Xsupport x86/reboot: Remove the RDC321X reboot quirk
I'm not sure about the RDC321X bits. Obviously the original 321x/861x/vortex86sx chips are obsolete and can be removed, but the product line is still actively developed by RDC and DM&P, and I suspect that some of the drivers are still used on 586tsc-class (vortex86dx, vortex86mx) and 686-class (vortex86dx3, vortex86ex) SoCs that do run modern kernels and get updates.
Quoted 2 lines
x86/cpu: Remove
CPU_SUP_UMC_32support x86/cpu: Remove TSC-lessCONFIG_M586support
I think Winchip6 (486-class, no tsc, no cx8) and Winchip3D (486-class, with tsc but no cx8) need to go as well then.
At this point, maybe we can consider removing CONFIG_X86_GENERIC and just always build kernels that work across a wide set of CPUs: Only CMOV and PAE still require a CPU with the hardware support, and X86_L1_CACHE_SHIFT needs to be at least 6 (64 byte) for compatibility, but everything else should just be a tuning option.
Arnd
Re: [RFC PATCH 0/15] x86: Remove support for TSC-less and CX8-less CPUs
Quoted 6 lines · Arnd Bergmann
Quoted 5 lines
x86/platform: Remove
CONFIG_X86_RDC321Xsupportarch/x86, gpio: RemoveGPIO_RDC321Xsupportarch/x86, watchdog: Remove theRDC321X_WDTwatchdog driverarch/x86, mfd: RemoveMFD_RDC321Xsupport x86/reboot: Remove the RDC321X reboot quirk
Quoted 7 lines
I'm not sure about the RDC321X bits. Obviously the original 321x/861x/vortex86sx chips are obsolete and can be removed, but the product line is still actively developed by RDC and DM&P, and I suspect that some of the drivers are still used on 586tsc-class (vortex86dx, vortex86mx) and 686-class (vortex86dx3, vortex86ex) SoCs that do run modern kernels and get updates.
So CONFIG_X86_RDC321X actively selects M486:
+++ b/arch/x86/Kconfig
config X86_RDC321X
bool "RDC R-321x SoC"
depends on X86_32
depends on X86_EXTENDED_PLATFORM
select M486
^^^^^^^^^^^
select X86_REBOOTFIXUPSBut indeed the other drivers are not dependent on M486, at least overtly:
arch/x86, mfd: Remove MFD_RDC321X support arch/x86, watchdog: Remove the RDC321X_WDT watchdog driver arch/x86, gpio: Remove GPIO_RDC321X support
Although the watchdog driver has this indirect dependency:
drivers/watchdog/Kconfig: depends on X86_RDC321X || COMPILE_TEST
But the 486 kernel would work on any 586/686 upgraded boards as well.
Anyway, I've dropped the mfd/watchdog/gpio removal patches, no harm in keeping these drivers, and I've switched the watchdog driver over to X86_32:
config RDC321X_WDT
tristate "RDC R-321x SoC watchdog"
depends on X86_32 || COMPILE_TESTThere's also no harm in keeping the southbridge reboot quirk I suppose, so I've dropped this as well:
x86/reboot: Remove the RDC321X reboot quirk
Quoted 5 lines
Quoted 2 lines
x86/cpu: Remove
CPU_SUP_UMC_32support x86/cpu: Remove TSC-lessCONFIG_M586supportI think Winchip6 (486-class, no tsc, no cx8) and Winchip3D (486-class, with tsc but no cx8) need to go as well then.
Okay, agreed, I've added this patch to the tree:
bf82539ad9f6 x86/cpu: Remove CONFIG_MWINCHIP3D/MWINCHIPC6
arch/x86/Kconfig.cpu | 28 ++++------------------------ arch/x86/Makefile_32.cpu | 2 -- arch/x86/include/asm/vermagic.h | 4 ---- 3 files changed, 4 insertions(+), 30 deletions(-)
Quoted 5 lines
At this point, maybe we can consider removing
CONFIG_X86_GENERICand just always build kernels that work across a wide set of CPUs: Only CMOV and PAE still require a CPU with the hardware support, andX86_L1_CACHE_SHIFTneeds to be at least 6 (64 byte) for compatibility, but everything else should just be a tuning option.
Agreed.
Thanks,
Ingo
Re: [RFC PATCH 0/15] x86: Remove support for TSC-less and CX8-less CPUs
Quoted 20 lines · Ingo Molnar
Quoted 6 lines · Arnd Bergmann
Quoted 5 lines
x86/platform: Remove
CONFIG_X86_RDC321Xsupportarch/x86, gpio: RemoveGPIO_RDC321Xsupportarch/x86, watchdog: Remove theRDC321X_WDTwatchdog driverarch/x86, mfd: RemoveMFD_RDC321Xsupport x86/reboot: Remove the RDC321X reboot quirkQuoted 7 lines
I'm not sure about the RDC321X bits. Obviously the original 321x/861x/vortex86sx chips are obsolete and can be removed, but the product line is still actively developed by RDC and DM&P, and I suspect that some of the drivers are still used on 586tsc-class (vortex86dx, vortex86mx) and 686-class (vortex86dx3, vortex86ex) SoCs that do run modern kernels and get updates.
So
CONFIG_X86_RDC321Xactively selects M486:+++ b/arch/x86/Kconfig
Quoted 7 lines
config
X86_RDC321Xbool "RDC R-321x SoC" depends on X86_32 depends on X86_EXTENDED_PLATFORM select M486 ^^^^^^^^^^^ select X86_REBOOTFIXUPS
Right, when the code got added, it was certainly for that specific chip, which I think is 486SX compatible.
The 'select M486' here doesn't actually do anything because Kconfig silently ignores 'select' for 'choice' symbols.
Quoted 7 lines
But indeed the other drivers are not dependent on M486, at least overtly:
arch/x86, mfd: RemoveMFD_RDC321Xsupportarch/x86, watchdog: Remove theRDC321X_WDTwatchdog driverarch/x86, gpio: RemoveGPIO_RDC321Xsupport
Quoted 6 lines
Although the watchdog driver has this indirect dependency:
drivers/watchdog/Kconfig: depends on X86_RDC321X || COMPILE_TESTAnyway, I've dropped the mfd/watchdog/gpio removal patches, no harm in keeping these drivers.
Thanks. We should still revisit all these separately and see which ones are used on more modern RDC/Vortex86 chips, as the relation between the brands isn't well documented.
I found an older lspci output from Xcore86MX/Vortex86MX showing that is uses an RDC R6021/R6036 bridge instead of R6020/R6030 on the RDC321x:
https://lore.kernel.org/all/[email protected]/
The Vortex86DX (585tsc compatible) datasheet in turn lists an R6021/R6031, which means the driver won't work out of the box, but it's probably not far off either if someone just adds the PCI ID.
Clearly nobody has done that so far, which would indicate that not a lot of people run vortex86 /and/ realize it's related to rdc321x.
Quoted 5 lines
and I've switched the watchdog driver over to
X86_32:config RDC321X_WDT tristate "RDC R-321x SoC watchdog" depends on X86_32 || COMPILE_TEST
How about 'CPU_SUP_VORTEX_32 || COMPILE_TEST'?
Quoted 4 lines
There's also no harm in keeping the southbridge reboot quirk I suppose, so I've dropped this as well:
x86/reboot: Remove the RDC321X reboot quirk
Right. Same thing here: the code probably still works on later R603x south bridges, but only triggers on the R6030 PCI ID that is not used on supported chips. Most likely nothing needs it.
Arnd
H. Peter Anvin
reviewerQuoted 64 lines · Linus Torvalds
Quoted 5 lines · Arnd Bergmann
Thanks for confirming. So a 486-targeted kernel still passes your tests on modern hardware if we force TSC and CX8 to be enabled, but the boot fails if the options are turned off in
Kconfig(though available in emulated hardware).I wouldn't expect CX8 to really matter - it causes us to generate extra code to pick one over the other, but on modern hardware we'll still always then dynamically pick the cmpxchg8b instruction.
Could it trigger bugs in our alternatives, or some miscompilation due to the extra complexity? Sure. But it does sound unlikely.
Quoted 3 lines
To be completely sure, you could re-run the same test with just one of these enabled, but I'm rather sure that the TSC is the root cause.
Agreed.
Particularly when the lockup is then in
timekeeping_notify()during the initial initcalls ->clocksource_select(), I'm pretty sure this is purely about TSC.That said, maybe the problem is in the watchdog logic, because
clocksource_done_booting()is what starts the watchdog thread .So it might be the watchdog code itself that then gets confused (because of some "don't use tsc" case that never gets any testing in real life) and triggers immediately - and then points the finger at the clocksource code only because that's what is still running.
Because
CONFIG_X86_TSCdoes cause some oddities: we end up still *using* the TSC for many things if the hardware supports it (which modern hardware obviously does), but then other things get disabled entirely.For example, this:
/*
* Boot-time check whether the TSCs are synchronized across
* all CPUs/cores:
*/
#ifdefCONFIG_X86_TSC
extern booltsc_store_and_check_tsc_adjust(bool bootcpu); extern voidtsc_verify_tsc_adjust(bool resume);
extern voidcheck_tsc_sync_target(void);
#else
static inline booltsc_store_and_check_tsc_adjust(bool bootcpu){
return false; }
static inline voidtsc_verify_tsc_adjust(bool resume){ } static inline voidcheck_tsc_sync_target(void){ } #endifSo that
tsc_store_and_check_tsc_adjust()thing etc never gets run, even though we actually *do* use TSC forget_cycles()and friends, because *that* code checks the runtime status too:Now, none of that should matter - because all *those* things are about details that simply aren't relevant for any of this case - but maybe there is some other situation that has similar "I'm actually using the TSC through
get_cycles(), but I didn't do some setup becauseX86_TSCwasn't on.."
Quoted 5 lines
I really get the feeling that it's time to leave i486 support behind. There's zero real reason for anybody to waste one second of development effort on this kind of issue.
Linus
Well, isn't the whole point that his patches remove the cx8 fallback code?
hi, Arnd,
Quoted 47 lines · Arnd Bergmann
Quoted 1 line · Oliver Sang
On Tue, Apr 22, 2025 at 12:16:33PM +0200, Arnd Bergmann wrote:
Cc: x86 and timekeeping maintainers, see
https://lore.kernel.org/lkml/[email protected]/
for the thread so far.Quoted 36 lines
Quoted 18 lines
Quoted 11 lines
[ 721.016779][ C0] hardirqs last disabled at (159506): sysvec_apic_timer_interrupt (arch/x86/kernel/apic/apic.c:1049) [ 721.016779][ C0] softirqs last enabled at (159174): handle_softirqs (kernel/softirq.c:408 kernel/softirq.c:589) [ 721.016779][ C0] softirqs last disabled at (159159): __do_softirq (kernel/softirq.c:596) [ 721.016779][ C0] CPU: 0 UID: 0 PID: 1 Comm: swapper Not tainted 6.14.0-rc3-00037-gf388f60ca904 #1 [ 721.016779][ C0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 [ 721.016779][ C0] EIP: timekeeping_notifyQuoted 1 line
(
kernel/time/timekeeping.c:1522)Timekeeping code could be related, I see that
CONFIG_X86_TSCis disabled for i486SX configurations, so even if a TSC is present in the emulated machine, it is not being used to measure time accurately.Quoted 5 lines
Quoted 1 line
-CONFIG_X86_CMPXCHG64=y
This could be another issue, if there is code that relies on the cx8/cmpxchg8b feature to be used. Since this is a non-SMP kernel, this is less likely to be the cause of the problem.
thanks a lot for all these details!
Quoted 5 lines
Can you try what happens when you enable the two options, either by changing
CONFIG_M486SXtoCONFIG_M586TSC, or with a patch like the one below? Note thatCONFIG_X86_CMPXCHG64recently got renamed toCONFIG_X86_CX8, but they are the exact same thing.I applied your patch directly upon
f388f60ca9(change forX86_CMPXCHG64instead ofX86_CX8as you metnioned), commit id isc1f7ef63239411313163a7b1bff654236f48351c...
Quoted 2 lines
by running same tests, now it backs to the clean status like
fc2d5cbe541032e7(parent off388f60ca9)
Quoted 8 lines
Thanks for confirming. So a 486-targeted kernel still passes your tests on modern hardware if we force TSC and CX8 to be enabled, but the boot fails if the options are turned off in
Kconfig(though available in emulated hardware).To be completely sure, you could re-run the same test with just one of these enabled, but I'm rather sure that the TSC is the root cause.
just FYI. we rerun the tests. if only enable X86_TSC, the config diff is
pkg/linux/i386-randconfig-r071-20250410/gcc-12/801597ddaae3bdc15546df3d0eba6e9e4e157b8d/.config 2025-04-25 14:24:49.488257697 +0800 renamed from pkg/linux/i386-randconfig-r071-20250410/gcc-12/f388f60ca9041a95c9b3f157d316ed7c8f297e44/.config 2025-04-15 15:41:17.009901645 +0800 +1 -0
@@ -351,6 +351,7 @@ CONFIG_X86_F00F_BUG=y CONFIG_X86_INVD_BUG=y CONFIG_X86_ALIGNMENT_16=y CONFIG_X86_INTEL_USERCOPY=y +CONFIG_X86_TSC=y CONFIG_X86_MINIMUM_CPU_FAMILY=4 CONFIG_IA32_FEAT_CTL=y CONFIG_X86_VMX_FEATURE_NAMES=y
the various issues still exists:fc2d5cbe541032e7 f388f60ca9041a95c9b3f157d31 801597ddaae3bdc15546df3d0eb ---------------- --------------------------- ---------------------------
fail:runs %reproduction fail:runs %reproduction fail:runs
| | | | |
:496 29% 145:494 31% 153:500 last_state.booting
:496 7% 35:494 14% 71:500 dmesg.BUG:kernel_hang_in_boot_stage
:496 0% :494 0% 1:500 dmesg.BUG:soft_lockup-CPU##stuck_for#s![kworker##:#]
:496 9% 45:494 5% 25:500 dmesg.BUG:soft_lockup-CPU##stuck_for#s![swapper:#]
:496 0% 1:494 0% :500 dmesg.EIP:__timer_delete_sync
:496 1% 5:494 1% 3:500 dmesg.EIP:_raw_spin_unlock_irq
:496 0% 2:494 0% 1:500 dmesg.EIP:_raw_spin_unlock_irqrestore
:496 0% 1:494 0% :500 dmesg.EIP:console_emit_next_record
:496 0% :494 0% 1:500 dmesg.EIP:finish_task_switch
:496 0% 1:494 1% 4:500 dmesg.EIP:handle_softirqs
:496 1% 3:494 1% 4:500 dmesg.EIP:lock_acquire
:496 0% 2:494 0% 2:500 dmesg.EIP:lock_release
:496 0% :494 0% 1:500 dmesg.EIP:preempt_schedule_thunk
:496 0% 1:494 0% :500 dmesg.EIP:queue_delayed_work_on
:496 0% :494 0% 1:500 dmesg.EIP:rmqueue_pcplist
:496 9% 45:494 5% 25:500 dmesg.EIP:timekeeping_notify
:496 3% 14:494 2% 8:500 dmesg.INFO:rcu_preempt_detected_stalls_on_CPUs/tasks
:496 0% :494 0% 1:500 dmesg.INFO:rcu_preempt_self-detected_stall_on_CPU
:496 6% 32:494 9% 46:500 dmesg.INFO:task_blocked_for_more_than#seconds
:496 9% 45:494 5% 26:500 dmesg.Kernel_panic-not_syncing:softlockup:hung_tasks
if only enable X86_CMPXCHG64:
pkg/linux/i386-randconfig-r071-20250410/gcc-12/dce28f73d4220df77c7fd6c41e854f1ef0af1d02/.config 2025-04-25 14:30:26.838037407 +0800 renamed from pkg/linux/i386-randconfig-r071-20250410/gcc-12/f388f60ca9041a95c9b3f157d316ed7c8f297e44/.config 2025-04-15 15:41:17.009901645 +0800 +3 -1
@@ -351,7 +351,8 @@ CONFIG_X86_F00F_BUG=y CONFIG_X86_INVD_BUG=y CONFIG_X86_ALIGNMENT_16=y CONFIG_X86_INTEL_USERCOPY=y -CONFIG_X86_MINIMUM_CPU_FAMILY=4 +CONFIG_X86_CMPXCHG64=y +CONFIG_X86_MINIMUM_CPU_FAMILY=5 CONFIG_IA32_FEAT_CTL=y CONFIG_X86_VMX_FEATURE_NAMES=y CONFIG_CPU_SUP_INTEL=y
@@ -5745,6 +5746,7 @@ CONFIG_GENERIC_NET_UTILS=y # CONFIG_PRIME_NUMBERS is not set CONFIG_RATIONAL=y CONFIG_GENERIC_IOMAP=y +CONFIG_ARCH_USE_CMPXCHG_LOCKREF=y CONFIG_ARCH_HAS_FAST_MULTIPLIER=y CONFIG_ARCH_USE_SYM_ANNOTATIONS=y
various issues gone:
fc2d5cbe541032e7 f388f60ca9041a95c9b3f157d31 dce28f73d4220df77c7fd6c41e8 ---------------- --------------------------- ---------------------------
fail:runs %reproduction fail:runs %reproduction fail:runs
| | | | |
:496 29% 145:494 0% :500 last_state.booting
:496 7% 35:494 0% :500 dmesg.BUG:kernel_hang_in_boot_stage
:496 9% 45:494 0% :500 dmesg.BUG:soft_lockup-CPU##stuck_for#s![swapper:#]
:496 0% 1:494 0% :500 dmesg.EIP:__timer_delete_sync
:496 1% 5:494 0% :500 dmesg.EIP:_raw_spin_unlock_irq
:496 0% 2:494 0% :500 dmesg.EIP:_raw_spin_unlock_irqrestore
:496 0% 1:494 0% :500 dmesg.EIP:console_emit_next_record
:496 0% 1:494 0% :500 dmesg.EIP:handle_softirqs
:496 1% 3:494 0% :500 dmesg.EIP:lock_acquire
:496 0% 2:494 0% :500 dmesg.EIP:lock_release
:496 0% 1:494 0% :500 dmesg.EIP:queue_delayed_work_on
:496 9% 45:494 0% :500 dmesg.EIP:timekeeping_notify
:496 3% 14:494 0% :500 dmesg.INFO:rcu_preempt_detected_stalls_on_CPUs/tasks
:496 6% 32:494 0% :500 dmesg.INFO:task_blocked_for_more_than#seconds
:496 9% 45:494 0% :500 dmesg.Kernel_panic-not_syncing:softlockup:hung_tasks
Quoted 28 lines
I tried reproducing the problem locally
with your .config on a qemu/tcg emulation running on an arm64 host, but this seems to run fine, including the rcutorture tests.Comparing my results with your log file, I see that your crash happens while changing the clocksource:
Your dmesg:
[ 92.548514][ T1] hpet0: 3 comparators, 64-bit 100.000000 MHz counter [ 721.016745][ C0] watchdog: BUG: soft lockup - CPU#0 stuck for 626s! [swapper:1]My dmesg:
[ 1.154511][ T1] hpet0: 3 comparators, 64-bit 100.000000 MHz counter [ 1.157896][ T1] clocksource: Switched to clocksource tsc-earlyThere are also clearly some differences between TCG and KVM in the handling of TSC, e.g. I get this warning from qemu itself for the SandyBridge CPU:
qemu-system-i386: warning: TCG doesn't support requested feature: CPUID.01H:ECX.tsc-deadline [bit 24]
I tried a few other variations, including KVM on an x86 laptop (using kvmclock or tsc-early clocksource), but none of them failed the way yours did.
Arnd
Quoted 12 lines · Oliver Sang
Quoted 9 lines · Arnd Bergmann
Thanks for confirming. So a 486-targeted kernel still passes your tests on modern hardware if we force TSC and CX8 to be enabled, but the boot fails if the options are turned off in
Kconfig(though available in emulated hardware).To be completely sure, you could re-run the same test with just one of these enabled, but I'm rather sure that the TSC is the root cause.
just FYI. we rerun the tests. if only enable
X86_TSC, the config diff is
...
Quoted 4 lines
the various issues still exists:
if only enable
X86_CMPXCHG64:
...
Quoted 1 line
various issues gone:
Interesting, so cx8 was indeed a problem. I would still assume that TSC caused the boot panic I cited, but it looks like CX8 caused all the other symptoms.
At the minimum this strengthens the case for the consensus of dropping support for all pre-586 cores.
Thanks a lot for confirming!
Arnd
Quoted 3 lines · Oliver Sang
We reran the tests. if only enable
X86_TSC, the various issues still exists. if only enableX86_CMPXCHG64, various issues gone.
Well, that's unexpected. I really didn't expect X86_CMPXCHG64 to make any difference, since we should still use the cmpxchg64 instruction, just with the alternative re-writing instead of directly.
Thanks for re-running the tests.
All the non-cmpxchg64 code sequences get replaced by the cmpxchg64 ones dynamically, so it all shouldn't matter one whit.
Except for during early boot. Because we do default to the old i386 sequences all the way *until* we do the alternatives replacement with the good cmpxchg64 ones.
It does change code generation, in that we have to have that alternative which now can be a call, so it's not a complete no-op, but I'm still surprised.
And except for not using CMPXCHG_LOCKREF at all, but that should be just a performance thing, and not noticeable during boot.
Hmm...
I'd love to understand why X86_CMPXCHG64 apparently matters, but I can't convince myself that it's worth really pursuing.
Linus
H. Peter Anvin
reviewerQuoted 25 lines · Linus Torvalds
Quoted 3 lines · Oliver Sang
We reran the tests. if only enable
X86_TSC, the various issues still exists. if only enableX86_CMPXCHG64, various issues gone.Well, that's unexpected. I really didn't expect
X86_CMPXCHG64to make any difference, since we should still use the cmpxchg64 instruction, just with the alternative re-writing instead of directly.Thanks for re-running the tests.
All the non-cmpxchg64 code sequences get replaced by the cmpxchg64 ones dynamically, so it all shouldn't matter one whit.
Except for during early boot. Because we do default to the old i386 sequences all the way *until* we do the alternatives replacement with the good cmpxchg64 ones.
It does change code generation, in that we have to have that alternative which now can be a call, so it's not a complete no-op, but I'm still surprised.
And except for not using
CMPXCHG_LOCKREFat all, but that should be just a performance thing, and not noticeable during boot.
Quoted 6 lines
Hmm...
I'd love to understand why
X86_CMPXCHG64apparently matters, but I can't convince myself that it's worth really pursuing.Linus
Sounds like the fallback stubs are subtly broken.
Personally I would have thought cmpxchg64 to be the biggest win in raising the baseline to the i586 ISA.