| Message ID | 1368699626.2618.183.camel@ThinkPad-T5421 (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Commit | cce606feb425093c8371089d392e336d186e125b |
| Delegated to: | Benjamin Herrenschmidt |
| Headers | show |
On Thu, 2013-05-16 at 18:20 +0800, Li Zhong wrote: > It seems following race is possible: > .../... > vdso_getcpu_init(); > #endif > - notify_cpu_starting(cpu); > - set_cpu_online(cpu, true); > /* Update sibling maps */ > base = cpu_first_thread_sibling(cpu); > for (i = 0; i < threads_per_core; i++) { > - if (cpu_is_offline(base + i)) > + if (cpu_is_offline(base + i) && (cpu != base + i)) > continue; > cpumask_set_cpu(cpu, cpu_sibling_mask(base + i)); > cpumask_set_cpu(base + i, cpu_sibling_mask(cpu)); > @@ -667,6 +665,10 @@ __cpuinit void start_secondary(void *unused) > } > of_node_put(l2_cache); > > + smp_wmb(); > + notify_cpu_starting(cpu); > + set_cpu_online(cpu, true); > + So we could have an online CPU with an empty sibling mask. Now we can have a sibling that isn't online ... Is that ok ? Or should we do a two pass mechanism: - Pass 1, set the new cpu siblings - set_cpu_online - Pass 2, set other CPU sibling of this cpu ? Cheers, Ben.
On 06/11/2013 12:30 PM, Benjamin Herrenschmidt wrote: > On Thu, 2013-05-16 at 18:20 +0800, Li Zhong wrote: >> It seems following race is possible: >> > > .../... > >> vdso_getcpu_init(); >> #endif >> - notify_cpu_starting(cpu); >> - set_cpu_online(cpu, true); >> /* Update sibling maps */ >> base = cpu_first_thread_sibling(cpu); >> for (i = 0; i < threads_per_core; i++) { >> - if (cpu_is_offline(base + i)) >> + if (cpu_is_offline(base + i) && (cpu != base + i)) >> continue; >> cpumask_set_cpu(cpu, cpu_sibling_mask(base + i)); >> cpumask_set_cpu(base + i, cpu_sibling_mask(cpu)); >> @@ -667,6 +665,10 @@ __cpuinit void start_secondary(void *unused) >> } >> of_node_put(l2_cache); >> >> + smp_wmb(); >> + notify_cpu_starting(cpu); >> + set_cpu_online(cpu, true); >> + > > So we could have an online CPU with an empty sibling mask. Now we can > have a sibling that isn't online ... Is that ok ? I think it is OK. We do the same thing on x86 as well - we set up the sibling links before calling notify_cpu_starting() and setting the cpu in the cpu_online_mask. In fact, there is even a comment explicitly noting that order: arch/x86/kernel/smpboot.c: 220 /* 221 * This must be done before setting cpu_online_mask 222 * or calling notify_cpu_starting. 223 */ 224 set_cpu_sibling_map(raw_smp_processor_id()); 225 wmb(); 226 227 notify_cpu_starting(cpuid); 228 229 /* 230 * Allow the master to continue. 231 */ 232 cpumask_set_cpu(cpuid, cpu_callin_mask); So I agree with Li Zhong's solution. [Arch-specific CPU hotplug code consolidation efforts such as [1] would have weeded out such nasty bugs.. I guess we should revive that patchset sometime soon.] Regards, Srivatsa S. Bhat [1]. https://lwn.net/Articles/500185/
On 05/16/2013 03:50 PM, Li Zhong wrote: > It seems following race is possible: > > cpu0 cpux > smp_init->cpu_up->_cpu_up > __cpu_up > kick_cpu(1) > ------------------------------------------------------------------------- > waiting online ... > ... notify CPU_STARTING > set cpux active > set cpux online > ------------------------------------------------------------------------- > finish waiting online > ... > sched_init_smp > init_sched_domains(cpu_active_mask) > build_sched_domains > set cpux sibling info > ------------------------------------------------------------------------- > [...] > This patch tries to move the sibling maps updating before > notify_cpu_starting() and cpu online, and a write barrier there to make > sure sibling maps are updated before active and online mask. > > Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com> > --- Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Regards, Srivatsa S. Bhat > arch/powerpc/kernel/smp.c | 8 +++++--- > 1 files changed, 5 insertions(+), 3 deletions(-) > > diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c > index ee7ac5e..c765937 100644 > --- a/arch/powerpc/kernel/smp.c > +++ b/arch/powerpc/kernel/smp.c > @@ -637,12 +637,10 @@ __cpuinit void start_secondary(void *unused) > > vdso_getcpu_init(); > #endif > - notify_cpu_starting(cpu); > - set_cpu_online(cpu, true); > /* Update sibling maps */ > base = cpu_first_thread_sibling(cpu); > for (i = 0; i < threads_per_core; i++) { > - if (cpu_is_offline(base + i)) > + if (cpu_is_offline(base + i) && (cpu != base + i)) > continue; > cpumask_set_cpu(cpu, cpu_sibling_mask(base + i)); > cpumask_set_cpu(base + i, cpu_sibling_mask(cpu)); > @@ -667,6 +665,10 @@ __cpuinit void start_secondary(void *unused) > } > of_node_put(l2_cache); > > + smp_wmb(); > + notify_cpu_starting(cpu); > + set_cpu_online(cpu, true); > + > local_irq_enable(); > > cpu_startup_entry(CPUHP_ONLINE); > > >
On Tue, 2013-06-11 at 14:03 +0530, Srivatsa S. Bhat wrote: > > So I agree with Li Zhong's solution. > > [Arch-specific CPU hotplug code consolidation efforts such as [1] would > have weeded out such nasty bugs.. I guess we should revive that patchset > sometime soon.] Thanks ! Cheers, Ben.
On Tue, 2013-06-11 at 14:03 +0530, Srivatsa S. Bhat wrote: > On 06/11/2013 12:30 PM, Benjamin Herrenschmidt wrote: > > On Thu, 2013-05-16 at 18:20 +0800, Li Zhong wrote: > >> It seems following race is possible: > >> > > > > .../... > > > >> vdso_getcpu_init(); > >> #endif > >> - notify_cpu_starting(cpu); > >> - set_cpu_online(cpu, true); > >> /* Update sibling maps */ > >> base = cpu_first_thread_sibling(cpu); > >> for (i = 0; i < threads_per_core; i++) { > >> - if (cpu_is_offline(base + i)) > >> + if (cpu_is_offline(base + i) && (cpu != base + i)) > >> continue; > >> cpumask_set_cpu(cpu, cpu_sibling_mask(base + i)); > >> cpumask_set_cpu(base + i, cpu_sibling_mask(cpu)); > >> @@ -667,6 +665,10 @@ __cpuinit void start_secondary(void *unused) > >> } > >> of_node_put(l2_cache); > >> > >> + smp_wmb(); > >> + notify_cpu_starting(cpu); > >> + set_cpu_online(cpu, true); > >> + > > > > So we could have an online CPU with an empty sibling mask. Now we can > > have a sibling that isn't online ... Is that ok ? > > I think it is OK. We do the same thing on x86 as well - we set up the > sibling links before calling notify_cpu_starting() and setting the cpu > in the cpu_online_mask. In fact, there is even a comment explicitly > noting that order: > > arch/x86/kernel/smpboot.c: > 220 /* > 221 * This must be done before setting cpu_online_mask > 222 * or calling notify_cpu_starting. > 223 */ > 224 set_cpu_sibling_map(raw_smp_processor_id()); > 225 wmb(); > 226 > 227 notify_cpu_starting(cpuid); > 228 > 229 /* > 230 * Allow the master to continue. > 231 */ > 232 cpumask_set_cpu(cpuid, cpu_callin_mask); > > > So I agree with Li Zhong's solution. > > [Arch-specific CPU hotplug code consolidation efforts such as [1] would > have weeded out such nasty bugs.. I guess we should revive that patchset > sometime soon.] > Thank you both for the review and comments. Good to know that it matches that of x86, and there is a patchset consolidating the code. With the patches in [1], it seems we only need the line to include the "to be onlined cpu" in this patch. Thanks, Zhong > Regards, > Srivatsa S. Bhat > > [1]. https://lwn.net/Articles/500185/ >
diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c index ee7ac5e..c765937 100644 --- a/arch/powerpc/kernel/smp.c +++ b/arch/powerpc/kernel/smp.c @@ -637,12 +637,10 @@ __cpuinit void start_secondary(void *unused) vdso_getcpu_init(); #endif - notify_cpu_starting(cpu); - set_cpu_online(cpu, true); /* Update sibling maps */ base = cpu_first_thread_sibling(cpu); for (i = 0; i < threads_per_core; i++) { - if (cpu_is_offline(base + i)) + if (cpu_is_offline(base + i) && (cpu != base + i)) continue; cpumask_set_cpu(cpu, cpu_sibling_mask(base + i)); cpumask_set_cpu(base + i, cpu_sibling_mask(cpu)); @@ -667,6 +665,10 @@ __cpuinit void start_secondary(void *unused) } of_node_put(l2_cache); + smp_wmb(); + notify_cpu_starting(cpu); + set_cpu_online(cpu, true); + local_irq_enable(); cpu_startup_entry(CPUHP_ONLINE);
It seems following race is possible: cpu0 cpux smp_init->cpu_up->_cpu_up __cpu_up kick_cpu(1) ------------------------------------------------------------------------- waiting online ... ... notify CPU_STARTING set cpux active set cpux online ------------------------------------------------------------------------- finish waiting online ... sched_init_smp init_sched_domains(cpu_active_mask) build_sched_domains set cpux sibling info ------------------------------------------------------------------------- Execution of cpu0 and cpux could be concurrent between two separator lines. So if the cpux sibling information was set too late (normally impossible, but could be triggered by adding some delay in start_secondary, after setting cpu online), build_sched_domains() running on cpu0 might see cpux active, with an empty sibling mask, then cause some bad address accessing like following: [ 0.099855] Unable to handle kernel paging request for data at address 0xc00000038518078f [ 0.099868] Faulting instruction address: 0xc0000000000b7a64 [ 0.099883] Oops: Kernel access of bad area, sig: 11 [#1] [ 0.099895] PREEMPT SMP NR_CPUS=16 DEBUG_PAGEALLOC NUMA pSeries [ 0.099922] Modules linked in: [ 0.099940] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.10.0-rc1-00120-gb973425-dirty #16 [ 0.099956] task: c0000001fed80000 ti: c0000001fed7c000 task.ti: c0000001fed7c000 [ 0.099971] NIP: c0000000000b7a64 LR: c0000000000b7a40 CTR: c0000000000b4934 [ 0.099985] REGS: c0000001fed7f760 TRAP: 0300 Not tainted (3.10.0-rc1-00120-gb973425-dirty) [ 0.099997] MSR: 8000000000009032 <SF,EE,ME,IR,DR,RI> CR: 24272828 XER: 20000003 [ 0.100045] SOFTE: 1 [ 0.100053] CFAR: c000000000445ee8 [ 0.100064] DAR: c00000038518078f, DSISR: 40000000 [ 0.100073] GPR00: 0000000000000080 c0000001fed7f9e0 c000000000c84d48 0000000000000010 GPR04: 0000000000000010 0000000000000000 c0000001fc55e090 0000000000000000 GPR08: ffffffffffffffff c000000000b80b30 c000000000c962d8 00000003845ffc5f GPR12: 0000000000000000 c00000000f33d000 c00000000000b9e4 0000000000000000 GPR16: 0000000000000000 0000000000000000 0000000000000001 0000000000000000 GPR20: c000000000ccf750 0000000000000000 c000000000c94d48 c0000001fc504000 GPR24: c0000001fc504000 c0000001fecef848 c000000000c94d48 c000000000ccf000 GPR28: c0000001fc522090 0000000000000010 c0000001fecef848 c0000001fed7fae0 [ 0.100293] NIP [c0000000000b7a64] .get_group+0x84/0xc4 [ 0.100307] LR [c0000000000b7a40] .get_group+0x60/0xc4 [ 0.100318] Call Trace: [ 0.100332] [c0000001fed7f9e0] [c0000000000dbce4] .lock_is_held+0xa8/0xd0 (unreliable) [ 0.100354] [c0000001fed7fa70] [c0000000000bf62c] .build_sched_domains+0x728/0xd14 [ 0.100375] [c0000001fed7fbe0] [c000000000af67bc] .sched_init_smp+0x4fc/0x654 [ 0.100394] [c0000001fed7fce0] [c000000000adce24] .kernel_init_freeable+0x17c/0x30c [ 0.100413] [c0000001fed7fdb0] [c00000000000ba08] .kernel_init+0x24/0x12c [ 0.100431] [c0000001fed7fe30] [c000000000009f74] .ret_from_kernel_thread+0x5c/0x68 [ 0.100445] Instruction dump: [ 0.100456] 38800010 38a00000 4838e3f5 60000000 7c6307b4 2fbf0000 419e0040 3d220001 [ 0.100496] 78601f24 39491590 e93e0008 7d6a002a <7d69582a> f97f0000 7d4a002a e93e0010 [ 0.100559] ---[ end trace 31fd0ba7d8756001 ]--- This patch tries to move the sibling maps updating before notify_cpu_starting() and cpu online, and a write barrier there to make sure sibling maps are updated before active and online mask. Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com> --- arch/powerpc/kernel/smp.c | 8 +++++--- 1 files changed, 5 insertions(+), 3 deletions(-)