Drop CONFIG_SCHED_MC
Helge Deller
maintainerPATCH 1/1 Drop CONFIG_SCHED_MC
Mikulas noticed that the parisc kernel crashes in sd_init() if CONFIG_SCHED_MC is enabled.
Multicore-scheduling is probably not very useful on parisc, so simply drop this option.
arch/parisc/Kconfig +0 -8
@@ -281,14 +281,6 @@ config SMP If you don't know what to do here, say N. -config SCHED_MC - bool "Multi-core scheduler support" - depends on GENERIC_ARCH_TOPOLOGY && PA8X00 - help - Multi-core scheduler support improves the CPU scheduler's decision - making when dealing with multi-core CPU chips at a cost of slightly - increased overhead in some places. If unsure say N here. - config IRQSTACKS bool "Use separate kernel stacks when processing interrupts" default y
arch/parisc/kernel/topology.c +0 -4
@@ -81,10 +81,6 @@ void store_cpu_topology(unsigned int cpuid) }static struct sched_domain_topology_level parisc_mc_topology[] = { -#ifdef CONFIG_SCHED_MC - { cpu_coregroup_mask, cpu_core_flags,SD_INIT_NAME (MC ) }, -#endif - { cpu_cpu_mask,SD_INIT_NAME (DIE ) }, {NULL , }, };
Quoted 8 lines · Helge Deller
Mikulas noticed that the parisc kernel crashes in
sd_init()ifCONFIG_SCHED_MCis enabled.
Multicore-scheduling is probably not very useful on parisc, so simply drop this option.Signed-off-by Helge Deller <[email protected]>Noticed-by Mikulas Patocka <[email protected]>Cc <[email protected]> # 5.18
Hi
I think that we should fix the root cause instead of trying to treat the symptoms.
Some more testing showed that:
in sd_init: tl->mask(cpu) returns an empty masktl->mask is cpu_coregroup_mask
in cpu_coregroup_mask: cpu_topology[cpu].core_sibling is an empty mask, that gets returned to sd_init
In arch/parisc/kernel/topology.c:init_cpu_topology is called before store_cpu_topology, but it depends on the variable dualcores_found being set by store_cpu_topology. Thus, it is not set.
store_cpu_topology returns if cpuid_topo->core_id != -1, but during boot, store_cpu_topology is called before reset_cpu_topology, thus the member "core_id" is uninitialized zero and store_cpu_tolopogy does nothing.
If these issues are addrssed, multicore scheduling will work.
Mikulas
Quoted 30 lines · Mikulas Patocka
Quoted 8 lines · Helge Deller
Mikulas noticed that the parisc kernel crashes in
sd_init()ifCONFIG_SCHED_MCis enabled.
Multicore-scheduling is probably not very useful on parisc, so simply drop this option.Signed-off-by Helge Deller <[email protected]>Noticed-by Mikulas Patocka <[email protected]>Cc <[email protected]> # 5.18Hi
I think that we should fix the root cause instead of trying to treat the symptoms.
Some more testing showed that:
in
sd_init:tl->mask(cpu)returns an empty masktl->maskiscpu_coregroup_mask
incpu_coregroup_mask: cpu_topology[cpu].core_sibling is an empty mask, that gets returned tosd_initIn
arch/parisc/kernel/topology.c:init_cpu_topologyis called beforestore_cpu_topology, but it depends on the variabledualcores_foundbeing set bystore_cpu_topology. Thus, it is not set.
Quoted 5 lines
store_cpu_topologyreturns ifcpuid_topo->core_id!= -1, but during boot,store_cpu_topologyis called beforereset_cpu_topology, thus the member "core_id" is uninitialized zero andstore_cpu_tolopogydoes nothing.If these issues are addrssed, multicore scheduling will work.
I've found that this fixes it.
Mikulas
--- arch/parisc/kernel/topology.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
arch/parisc/kernel/topology.c +2 -1
@@ -95,7 +95,8 @@ static struct sched_domain_topology_leve*/ void __init init_cpu_topology (void ) { +reset_cpu_topology ();/* Set scheduler topology descriptor */ -if (dualcores_found) +/*if (dualcores_found)*/ set_sched_topology (parisc_mc_topology); }
PATCH 1/1 parisc: fix a crash with multicore scheduler
With the kernel 5.18, the system will hang on boot if it is compiled with CONFIG_SCHED_MC. The last printed message is "Brought up 1 node, 1 CPU".
The crash happens in sd_inittl->mask (which is cpu_coregroup_mask) returns an empty mask. This happens
because cpu_topology[0].core_sibling is empty.
Consequently, sd_span is set to an empty masksd_id = cpumask_first(sd_span) sets sd_id == NR_CPUS (because the mask is
empty)sd->shared = *per_cpu_ptr(sdd->sds, sd_id); sets sd->shared to NULL
because sd_id is out of rangeatomic_inc(&sd->shared->ref); crashes without printing anything
We can fix it by calling reset_cpu_topology() from init_cpu_topology() - this will initialize the sibling masks on CPUs, so that they're not empty.
This patch also removes the variable "dualcores_found", it is useless, because during boot, init_cpu_topology is called before store_cpu_topology. Thus, set_sched_topology(parisc_mc_topology) is never called. We don't need to call it at all because default_topology in kernel/sched/topology.c contains the same items as parisc_mc_topology.
--- arch/parisc/kernel/topology.c | 16 +--------------- 1 file changed, 1 insertion(+), 15 deletions(-)
arch/parisc/kernel/topology.c +1 -15
@@ -20,8 +20,6 @@ static DEFINE_PER_CPU(struct cpu, cpu_devices); -static int dualcores_found; -/* * store_cpu_topology is called at boot when only one cpu is running * and with the mutex cpu_hotplug.lock locked, when several cpus have booted,
@@ -60,7 +58,6 @@ void store_cpu_topology(unsigned int cpuif (p->cpu_loc) { cpuid_topo->core_id++; cpuid_topo->package_id = cpu_topology[cpu].package_id; - dualcores_found =1 ;continue ; } }
@@ -80,22 +77,11 @@ void store_cpu_topology(unsigned int cpu cpu_topology[cpuid].package_id); } -static struct sched_domain_topology_level parisc_mc_topology[] = { -#ifdef CONFIG_SCHED_MC - { cpu_coregroup_mask, cpu_core_flags,SD_INIT_NAME (MC ) }, -#endif - - { cpu_cpu_mask,SD_INIT_NAME (DIE ) }, - {NULL , }, -}; -/* * init_cpu_topology is called at boot when only one cpu is running * which prevent simultaneous write access to cpu_topology array */ void __init init_cpu_topology (void ) { -/* Set scheduler topology descriptor */ -if (dualcores_found) -set_sched_topology (parisc_mc_topology); +reset_cpu_topology (); }
PATCH v2 1/1 parisc: fix a crash with multicore scheduler
With the kernel 5.18, the system will hang on boot if it is compiled with CONFIG_SCHED_MC. The last printed message is "Brought up 1 node, 1 CPU".
The crash happens in sd_inittl->mask (which is cpu_coregroup_mask) returns an empty mask. This happens
because cpu_topology[0].core_sibling is empty.
Consequently, sd_span is set to an empty masksd_id = cpumask_first(sd_span) sets sd_id == NR_CPUS (because the mask is
empty)sd->shared = *per_cpu_ptr(sdd->sds, sd_id); sets sd->shared to NULL
because sd_id is out of rangeatomic_inc(&sd->shared->ref); crashes without printing anything
We can fix it by calling reset_cpu_topology() from init_cpu_topology() - this will initialize the sibling masks on CPUs, so that they're not empty.
This patch also removes the variable "dualcores_found", it is useless, because during boot, init_cpu_topology is called before store_cpu_topology. Thus, set_sched_topology(parisc_mc_topology) is never called. We don't need to call it at all because default_topology in kernel/sched/topology.c contains the same items as parisc_mc_topology.
Note that we should not call store_cpu_topology() from init_per_cpu() because it is called too early in the kernel initialization process and it results in the message "Failure to register CPU0 device". Before this patch, store_cpu_topology() would exit immediatelly because cpuid_topo->core id was uninitialized and it was 0.
--- arch/parisc/kernel/processor.c | 2 -- arch/parisc/kernel/topology.c | 16 +--------------- 2 files changed, 1 insertion(+), 17 deletions(-)
arch/parisc/kernel/topology.c +1 -15
@@ -20,8 +20,6 @@ static DEFINE_PER_CPU(struct cpu, cpu_devices); -static int dualcores_found; -/* * store_cpu_topology is called at boot when only one cpu is running * and with the mutex cpu_hotplug.lock locked, when several cpus have booted,
@@ -60,7 +58,6 @@ void store_cpu_topology(unsigned int cpuif (p->cpu_loc) { cpuid_topo->core_id++; cpuid_topo->package_id = cpu_topology[cpu].package_id; - dualcores_found =1 ;continue ; } }
@@ -80,22 +77,11 @@ void store_cpu_topology(unsigned int cpu cpu_topology[cpuid].package_id); } -static struct sched_domain_topology_level parisc_mc_topology[] = { -#ifdef CONFIG_SCHED_MC - { cpu_coregroup_mask, cpu_core_flags,SD_INIT_NAME (MC ) }, -#endif - - { cpu_cpu_mask,SD_INIT_NAME (DIE ) }, - {NULL , }, -}; -/* * init_cpu_topology is called at boot when only one cpu is running * which prevent simultaneous write access to cpu_topology array */ void __init init_cpu_topology (void ) { -/* Set scheduler topology descriptor */ -if (dualcores_found) -set_sched_topology (parisc_mc_topology); +reset_cpu_topology (); }
arch/parisc/kernel/processor.c +0 -2
@@ -327,8 +327,6 @@ int init_per_cpu(int cpunum)set_firmware_width (); ret =pdc_coproc_cfg (&coproc_cfg); -store_cpu_topology (cpunum); -if (ret >=0 && coproc_cfg.ccr_functional) {mtctl (coproc_cfg.ccr_functional,10 );/* 10 == Coprocessor Control Reg */
Helge Deller
maintainerRe: [PATCH v2] parisc: fix a crash with multicore scheduler
Hi Mikulas,
Quoted 22 lines · Mikulas Patocka
With the kernel 5.18, the system will hang on boot if it is compiled with
CONFIG_SCHED_MC. The last printed message is "Brought up 1 node, 1 CPU".The crash happens in
sd_inittl->mask(which iscpu_coregroup_mask) returns an empty mask. This happens
because cpu_topology[0].core_sibling is empty.
Consequently,sd_spanis set to an empty masksd_id=cpumask_first(sd_span)setssd_id==NR_CPUS(because the mask is
empty)sd->shared= *per_cpu_ptr(sdd->sds,sd_id); setssd->sharedto NULL
becausesd_idis out of rangeatomic_inc(&sd->shared->ref); crashes without printing anythingWe can fix it by calling
reset_cpu_topology()frominit_cpu_topology()- this will initialize the sibling masks on CPUs, so that they're not empty.This patch also removes the variable "
dualcores_found", it is useless, because during boot,init_cpu_topologyis called beforestore_cpu_topology. Thus,set_sched_topology(parisc_mc_topology)is never called. We don't need to call it at all becausedefault_topologyinkernel/sched/topology.ccontains the same items asparisc_mc_topology.
Quoted 8 lines
Note that we should not call
store_cpu_topology()frominit_per_cpu()because it is called too early in the kernel initialization process and it results in the message "Failure to register CPU0 device". Before this patch,store_cpu_topology()would exit immediatelly becausecpuid_topo->coreid was uninitialized and it was 0.Signed-off-by Mikulas Patocka <[email protected]>Cc [email protected] # v5.18
Thanks a lot !!!
It took me some time to test it, but it looks good and boots on all of my machines so far. I was curious if 32-bit kernels still work since that was one of the issues with the older patches...
With your patch we can drop the "config SCHED_MC" entry from arch/parisc/Kconfig as well.
Will you respin, or should I simply add this to your patch?
Helge
Quoted 65 lines
--- arch/parisc/kernel/processor.c | 2 -- arch/parisc/kernel/topology.c | 16 +--------------- 2 files changed, 1 insertion(+), 17 deletions(-)Index:
linux-2.6/arch/parisc/kernel/topology.c
===================================================================arch/parisc/kernel/topology.c +1
-15@@ -20,8 +20,6 @@ static DEFINE_PER_CPU(struct cpu, cpu_devices); -static int dualcores_found; -/* * store_cpu_topology is called at boot when only one cpu is running * and with the mutex cpu_hotplug.lock locked, when several cpus have booted, @@ -60,7 +58,6 @@ void store_cpu_topology(unsigned int cpuif (p->cpu_loc) { cpuid_topo->core_id++; cpuid_topo->package_id = cpu_topology[cpu].package_id; - dualcores_found =1 ;continue ; } }@@ -80,22 +77,11 @@ void store_cpu_topology(unsigned int cpu cpu_topology[cpuid].package_id); } -static struct sched_domain_topology_level parisc_mc_topology[] = { -#ifdef CONFIG_SCHED_MC - { cpu_coregroup_mask, cpu_core_flags,SD_INIT_NAME (MC ) }, -#endif - - { cpu_cpu_mask,SD_INIT_NAME (DIE ) }, - {NULL , }, -}; -/* * init_cpu_topology is called at boot when only one cpu is running * which prevent simultaneous write access to cpu_topology array */ void __init init_cpu_topology (void ) { -/* Set scheduler topology descriptor */ -if (dualcores_found) -set_sched_topology (parisc_mc_topology); +reset_cpu_topology (); }Index:
linux-2.6/arch/parisc/kernel/processor.c
===================================================================arch/parisc/kernel/processor.c +0
-2@@ -327,8 +327,6 @@ int init_per_cpu(int cpunum)set_firmware_width (); ret =pdc_coproc_cfg (&coproc_cfg); -store_cpu_topology (cpunum); -if (ret >=0 && coproc_cfg.ccr_functional) {mtctl (coproc_cfg.ccr_functional,10 );/* 10 == Coprocessor Control Reg */
Re: [PATCH v2] parisc: fix a crash with multicore scheduler
Quoted 8 lines · Helge Deller
Hi Mikulas,
Thanks a lot !!!
It took me some time to test it, but it looks good and boots on all of my machines so far. I was curious if 32-bit kernels still work since that was one of the issues with the older patches...
Quoted 5 lines
With your patch we can drop the "config
SCHED_MC" entry fromarch/parisc/Kconfigas well.
Will you respin, or should I simply add this to your patch?Helge
I think that we don't have to drop "config SCHED_MC". It is used in kernel/sched/topology.c to select the multicore-aware scheduler. There is no reason why the multicore scheduler would not work on parisc.
Mikulas