* [PATCH 0/4] blackfin: convet cpumask apis
@ 2011-04-25 9:27 KOSAKI Motohiro
2011-04-25 9:28 ` [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro
` (5 more replies)
0 siblings, 6 replies; 17+ messages in thread
From: KOSAKI Motohiro @ 2011-04-25 9:27 UTC (permalink / raw)
To: Michael Hennerich, device-driver-devel, LKML, Andrew Morton
Cc: kosaki.motohiro
Rusty Russle introduced a lot of cpumask related APIs. and He gone even
though the work is unfinished. This patch series is sequel of his work.
No functional change. only api change.
note: I've confirmed only cross-compile build. so, I hope blackfin
developers see the code.
KOSAKI Motohiro (4):
blackfin: remove unused function
blackfin: don't touch cpu_possible_map and cpu_present_map directly
blackfin: don't touch task->cpus_allowed directly
blackfin: convert old cpumask API to new one
arch/blackfin/kernel/nmi.c | 8 ++--
arch/blackfin/kernel/process.c | 6 +--
arch/blackfin/kernel/setup.c | 4 +-
arch/blackfin/mach-bf561/include/mach/smp.h | 2 -
arch/blackfin/mach-bf561/smp.c | 27 ++++-------
arch/blackfin/mach-common/dpmc.c | 7 ++-
arch/blackfin/mach-common/smp.c | 27 ++++++-----
^ permalink raw reply [flat|nested] 17+ messages in thread* [PATCH 1/4] blackfin: remove unused function 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro @ 2011-04-25 9:28 ` KOSAKI Motohiro 2011-04-25 9:29 ` [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro ` (4 subsequent siblings) 5 siblings, 0 replies; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-25 9:28 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, Michael Hennerich, device-driver-devel, LKML, Andrew Morton platform_send_ipi() has old call-by-value cpumask_t fashon and it's unused. Then, this patch removes it. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Michael Hennerich <michael.hennerich@analog.com> Cc: device-driver-devel@blackfin.uclinux.org --- arch/blackfin/mach-bf561/include/mach/smp.h | 2 -- arch/blackfin/mach-bf561/smp.c | 13 ------------- 2 files changed, 0 insertions(+), 15 deletions(-) diff --git a/arch/blackfin/mach-bf561/include/mach/smp.h b/arch/blackfin/mach-bf561/include/mach/smp.h index 346c605..36e2ac5 100644 --- a/arch/blackfin/mach-bf561/include/mach/smp.h +++ b/arch/blackfin/mach-bf561/include/mach/smp.h @@ -21,8 +21,6 @@ void platform_secondary_init(unsigned int cpu); void platform_request_ipi(int irq, /*irq_handler_t*/ void *handler); -void platform_send_ipi(cpumask_t callmap, int irq); - void platform_send_ipi_cpu(unsigned int cpu, int irq); void platform_clear_ipi(unsigned int cpu, int irq); diff --git a/arch/blackfin/mach-bf561/smp.c b/arch/blackfin/mach-bf561/smp.c index 7b07740..630e2c2 100644 --- a/arch/blackfin/mach-bf561/smp.c +++ b/arch/blackfin/mach-bf561/smp.c @@ -116,19 +116,6 @@ void __init platform_request_ipi(int irq, void *handler) panic("Cannot request %s for IPI service", name); } -void platform_send_ipi(cpumask_t callmap, int irq) -{ - unsigned int cpu; - int offset = (irq == IRQ_SUPPLE_0) ? 6 : 8; - - for_each_cpu_mask(cpu, callmap) { - BUG_ON(cpu >= 2); - SSYNC(); - bfin_write_SICB_SYSCR(bfin_read_SICB_SYSCR() | (1 << (offset + cpu))); - SSYNC(); - } -} - void platform_send_ipi_cpu(unsigned int cpu, int irq) { int offset = (irq == IRQ_SUPPLE_0) ? 6 : 8; -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro 2011-04-25 9:28 ` [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro @ 2011-04-25 9:29 ` KOSAKI Motohiro 2011-04-25 9:31 ` [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro ` (3 subsequent siblings) 5 siblings, 0 replies; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-25 9:29 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, Michael Hennerich, device-driver-devel, LKML, Andrew Morton We plan to remove cpu_possible_map and cpu_present_map later and we have proper init_cpu_possible() and init_cpu_present() APIs. Therefore this patch rewrites platform_init_cpus and platform_prepare_cpus by their APIs. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Michael Hennerich <michael.hennerich@analog.com> Cc: device-driver-devel@blackfin.uclinux.org --- arch/blackfin/mach-bf561/smp.c | 14 ++++++++++---- 1 files changed, 10 insertions(+), 4 deletions(-) diff --git a/arch/blackfin/mach-bf561/smp.c b/arch/blackfin/mach-bf561/smp.c index 630e2c2..2fca469 100644 --- a/arch/blackfin/mach-bf561/smp.c +++ b/arch/blackfin/mach-bf561/smp.c @@ -24,17 +24,23 @@ static DEFINE_SPINLOCK(boot_lock); void __init platform_init_cpus(void) { - cpu_set(0, cpu_possible_map); /* CoreA */ - cpu_set(1, cpu_possible_map); /* CoreB */ + struct cpumask mask; + + cpumask_set_cpu(0, &mask); /* CoreA */ + cpumask_set_cpu(1, &mask); /* CoreB */ + init_cpu_possible(&mask); } void __init platform_prepare_cpus(unsigned int max_cpus) { + struct cpumask mask; + bfin_relocate_coreb_l1_mem(); /* Both cores ought to be present on a bf561! */ - cpu_set(0, cpu_present_map); /* CoreA */ - cpu_set(1, cpu_present_map); /* CoreB */ + cpumask_set_cpu(0, &mask); /* CoreA */ + cpumask_set_cpu(1, &mask); /* CoreB */ + init_cpu_present(&mask); } int __init setup_profiling_timer(unsigned int multiplier) /* not supported */ -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro 2011-04-25 9:28 ` [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro 2011-04-25 9:29 ` [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro @ 2011-04-25 9:31 ` KOSAKI Motohiro 2011-04-25 9:32 ` [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro ` (2 subsequent siblings) 5 siblings, 0 replies; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-25 9:31 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, Michael Hennerich, device-driver-devel, LKML, Andrew Morton Every callter (except kthread_bind) should use proper set_cpus_allowed_ptr() APIs. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Michael Hennerich <michael.hennerich@analog.com> Cc: device-driver-devel@blackfin.uclinux.org --- I'm curious why this mysterious code is necessary. Why sys_clone() restrict allowed cpus automatically and why don't it restore the restriction when do_fork() is finished. arch/blackfin/kernel/process.c | 6 ++---- 1 files changed, 2 insertions(+), 4 deletions(-) diff --git a/arch/blackfin/kernel/process.c b/arch/blackfin/kernel/process.c index b407bc8..6a660fa 100644 --- a/arch/blackfin/kernel/process.c +++ b/arch/blackfin/kernel/process.c @@ -171,10 +171,8 @@ asmlinkage int bfin_clone(struct pt_regs *regs) unsigned long newsp; #ifdef __ARCH_SYNC_CORE_DCACHE - if (current->rt.nr_cpus_allowed == num_possible_cpus()) { - current->cpus_allowed = cpumask_of_cpu(smp_processor_id()); - current->rt.nr_cpus_allowed = 1; - } + if (current->rt.nr_cpus_allowed == num_possible_cpus()) + set_cpus_allowed_ptr(current, cpumask_of(smp_processor_id())); #endif /* syscall2 puts clone_flags in r0 and usp in r1 */ -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 4/4] blackfin: convert old cpumask API to new one 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro ` (2 preceding siblings ...) 2011-04-25 9:31 ` [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro @ 2011-04-25 9:32 ` KOSAKI Motohiro 2011-04-25 18:11 ` [PATCH 0/4] blackfin: convet cpumask apis Mike Frysinger 2011-04-26 18:26 ` [PATCH 0/4] blackfin: convet cpumask apis Thiago Farina 5 siblings, 0 replies; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-25 9:32 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, Michael Hennerich, device-driver-devel, LKML, Andrew Morton old cpu_xxx() APIs is planned to removed later. then, converted. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Michael Hennerich <michael.hennerich@analog.com> Cc: device-driver-devel@blackfin.uclinux.org --- arch/blackfin/kernel/nmi.c | 8 ++++---- arch/blackfin/kernel/setup.c | 4 ++-- arch/blackfin/mach-common/dpmc.c | 7 ++++--- arch/blackfin/mach-common/smp.c | 27 ++++++++++++++------------- 4 files changed, 24 insertions(+), 22 deletions(-) diff --git a/arch/blackfin/kernel/nmi.c b/arch/blackfin/kernel/nmi.c index 0b5f72f..7af8b70 100644 --- a/arch/blackfin/kernel/nmi.c +++ b/arch/blackfin/kernel/nmi.c @@ -145,16 +145,16 @@ int check_nmi_wdt_touched(void) { unsigned int this_cpu = smp_processor_id(); unsigned int cpu; + cpumask_t mask; - cpumask_t mask = cpu_online_map; - + cpumask_copy(&mask, cpu_online_mask); if (!atomic_read(&nmi_touched[this_cpu])) return 0; atomic_set(&nmi_touched[this_cpu], 0); - cpu_clear(this_cpu, mask); - for_each_cpu_mask(cpu, mask) { + cpumask_clear_cpu(this_cpu, &mask); + for_each_cpu(cpu,&mask) { invalidate_dcache_range((unsigned long)(&nmi_touched[cpu]), (unsigned long)(&nmi_touched[cpu])); if (!atomic_read(&nmi_touched[cpu])) diff --git a/arch/blackfin/kernel/setup.c b/arch/blackfin/kernel/setup.c index 805c613..c00c87b 100644 --- a/arch/blackfin/kernel/setup.c +++ b/arch/blackfin/kernel/setup.c @@ -1326,7 +1326,7 @@ static int show_cpuinfo(struct seq_file *m, void *v) static void *c_start(struct seq_file *m, loff_t *pos) { if (*pos == 0) - *pos = first_cpu(cpu_online_map); + *pos = cpumask_first(cpu_online_mask); if (*pos >= num_online_cpus()) return NULL; @@ -1335,7 +1335,7 @@ static void *c_start(struct seq_file *m, loff_t *pos) static void *c_next(struct seq_file *m, void *v, loff_t *pos) { - *pos = next_cpu(*pos, cpu_online_map); + *pos = cpumask_next(*pos, cpu_online_mask); return c_start(m, pos); } diff --git a/arch/blackfin/mach-common/dpmc.c b/arch/blackfin/mach-common/dpmc.c index 382099f..6ea2921 100644 --- a/arch/blackfin/mach-common/dpmc.c +++ b/arch/blackfin/mach-common/dpmc.c @@ -88,10 +88,11 @@ static void bfin_wakeup_cpu(void) { unsigned int cpu; unsigned int this_cpu = smp_processor_id(); - cpumask_t mask = cpu_online_map; + cpumask_t mask; - cpu_clear(this_cpu, mask); - for_each_cpu_mask(cpu, mask) + cpumask_copy(&mask, cpu_online_mask); + cpumask_clear_cpu(this_cpu, &mask); + for_each_cpu(cpu, &mask) platform_send_ipi_cpu(cpu, IRQ_SUPPLE_0); } diff --git a/arch/blackfin/mach-common/smp.c b/arch/blackfin/mach-common/smp.c index 1fbd94c..f430bba 100644 --- a/arch/blackfin/mach-common/smp.c +++ b/arch/blackfin/mach-common/smp.c @@ -96,7 +96,7 @@ static void ipi_cpu_stop(unsigned int cpu) dump_stack(); spin_unlock(&stop_lock); - cpu_clear(cpu, cpu_online_map); + set_cpu_online(cpu, false); local_irq_disable(); @@ -146,7 +146,7 @@ static void ipi_call_function(unsigned int cpu, struct ipi_message *msg) */ resync_core_dcache(); #endif - cpu_clear(cpu, *msg->call_struct.waitmask); + cpumask_clear_cpu(cpu, msg->call_struct.waitmask); } } @@ -222,9 +222,10 @@ static inline void smp_send_message(cpumask_t callmap, unsigned long type, struct ipi_message_queue *msg_queue; struct ipi_message *msg; unsigned long flags, next_msg; - cpumask_t waitmask = callmap; /* waitmask is shared by all cpus */ + cpumask_t waitmask; /* waitmask is shared by all cpus */ - for_each_cpu_mask(cpu, callmap) { + cpumask_copy(&waitmask, &callmap); + for_each_cpu(cpu, &callmap) { msg_queue = &per_cpu(ipi_msg_queue, cpu); spin_lock_irqsave(&msg_queue->lock, flags); if (msg_queue->count < BFIN_IPI_MSGQ_LEN) { @@ -246,7 +247,7 @@ static inline void smp_send_message(cpumask_t callmap, unsigned long type, } if (wait) { - while (!cpus_empty(waitmask)) + while (!cpumask_empty(&waitmask)) blackfin_dcache_invalidate_range( (unsigned long)(&waitmask), (unsigned long)(&waitmask)); @@ -265,9 +266,9 @@ int smp_call_function(void (*func)(void *info), void *info, int wait) cpumask_t callmap; preempt_disable(); - callmap = cpu_online_map; - cpu_clear(smp_processor_id(), callmap); - if (!cpus_empty(callmap)) + cpumask_copy(&callmap, cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), &callmap); + if (!cpumask_empty(&callmap)) smp_send_message(callmap, BFIN_IPI_CALL_FUNC, func, info, wait); preempt_enable(); @@ -284,8 +285,8 @@ int smp_call_function_single(int cpuid, void (*func) (void *info), void *info, if (cpu_is_offline(cpu)) return 0; - cpus_clear(callmap); - cpu_set(cpu, callmap); + cpumask_clear(&callmap); + cpumask_set_cpu(cpu, &callmap); smp_send_message(callmap, BFIN_IPI_CALL_FUNC, func, info, wait); @@ -308,9 +309,9 @@ void smp_send_stop(void) cpumask_t callmap; preempt_disable(); - callmap = cpu_online_map; - cpu_clear(smp_processor_id(), callmap); - if (!cpus_empty(callmap)) + cpumask_copy(&callmap, cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), &callmap); + if (!cpumask_empty(&callmap)) smp_send_message(callmap, BFIN_IPI_CPU_STOP, NULL, NULL, 0); preempt_enable(); -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH 0/4] blackfin: convet cpumask apis 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro ` (3 preceding siblings ...) 2011-04-25 9:32 ` [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro @ 2011-04-25 18:11 ` Mike Frysinger 2011-04-25 18:18 ` Mike Frysinger 2011-04-26 18:26 ` [PATCH 0/4] blackfin: convet cpumask apis Thiago Farina 5 siblings, 1 reply; 17+ messages in thread From: Mike Frysinger @ 2011-04-25 18:11 UTC (permalink / raw) To: KOSAKI Motohiro Cc: Michael Hennerich, device-driver-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 05:27, KOSAKI Motohiro wrote: > Rusty Russle introduced a lot of cpumask related APIs. and He gone even > though the work is unfinished. This patch series is sequel of his work. > > No functional change. only api change. > > note: I've confirmed only cross-compile build. so, I hope blackfin > developers see the code. fyi, device-driver-devel@b.u.o is for general device drivers while uclinux-dist-devel@b.u.o is for Blackfin arch work -mike ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 0/4] blackfin: convet cpumask apis 2011-04-25 18:11 ` [PATCH 0/4] blackfin: convet cpumask apis Mike Frysinger @ 2011-04-25 18:18 ` Mike Frysinger 2011-04-26 1:49 ` KOSAKI Motohiro 0 siblings, 1 reply; 17+ messages in thread From: Mike Frysinger @ 2011-04-25 18:18 UTC (permalink / raw) To: KOSAKI Motohiro Cc: Michael Hennerich, device-driver-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 14:11, Mike Frysinger wrote: > On Mon, Apr 25, 2011 at 05:27, KOSAKI Motohiro wrote: >> Rusty Russle introduced a lot of cpumask related APIs. and He gone even >> though the work is unfinished. This patch series is sequel of his work. >> >> No functional change. only api change. >> >> note: I've confirmed only cross-compile build. so, I hope blackfin >> developers see the code. > > fyi, device-driver-devel@b.u.o is for general device drivers while > uclinux-dist-devel@b.u.o is for Blackfin arch work oh, and i guess you typo-ed the list name since it's "device-drivers-devel" ;). probably better to resend the series and not cc Michael (since he doesnt watch over Blackfin anymore). -mike ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 0/4] blackfin: convet cpumask apis 2011-04-25 18:18 ` Mike Frysinger @ 2011-04-26 1:49 ` KOSAKI Motohiro 2011-04-26 1:53 ` (resend) [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro 0 siblings, 1 reply; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-26 1:49 UTC (permalink / raw) To: Mike Frysinger Cc: kosaki.motohiro, Michael Hennerich, device-driver-devel, LKML, Andrew Morton > On Mon, Apr 25, 2011 at 14:11, Mike Frysinger wrote: > > On Mon, Apr 25, 2011 at 05:27, KOSAKI Motohiro wrote: > >> Rusty Russle introduced a lot of cpumask related APIs. and He gone even > >> though the work is unfinished. This patch series is sequel of his work. > >> > >> No functional change. only api change. > >> > >> note: I've confirmed only cross-compile build. so, I hope blackfin > >> developers see the code. > > > > fyi, device-driver-devel@b.u.o is for general device drivers while > > uclinux-dist-devel@b.u.o is for Blackfin arch work > > oh, and i guess you typo-ed the list name since it's > "device-drivers-devel" ;). probably better to resend the series and > not cc Michael (since he doesnt watch over Blackfin anymore). I'm sorry. Will do. ^ permalink raw reply [flat|nested] 17+ messages in thread
* (resend) [PATCH 1/4] blackfin: remove unused function 2011-04-26 1:49 ` KOSAKI Motohiro @ 2011-04-26 1:53 ` KOSAKI Motohiro 2011-04-26 1:55 ` (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro ` (3 more replies) 0 siblings, 4 replies; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-26 1:53 UTC (permalink / raw) To: uclinux-dist-devel; +Cc: kosaki.motohiro, Mike Frysinger, LKML, Andrew Morton platform_send_ipi() has old call-by-value cpumask_t fashon and it's unused. Then, this patch removes it. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Mike Frysinger <vapier@gentoo.org> Cc: uclinux-dist-devel@blackfin.uclinux.org --- arch/blackfin/mach-bf561/include/mach/smp.h | 2 -- arch/blackfin/mach-bf561/smp.c | 13 ------------- 2 files changed, 0 insertions(+), 15 deletions(-) I sent this parch to wrong address (device-driver-devel@blackfin.uclinux.org). thus now I'm resending it. diff --git a/arch/blackfin/mach-bf561/include/mach/smp.h b/arch/blackfin/mach-bf561/include/mach/smp.h index 346c605..36e2ac5 100644 --- a/arch/blackfin/mach-bf561/include/mach/smp.h +++ b/arch/blackfin/mach-bf561/include/mach/smp.h @@ -21,8 +21,6 @@ void platform_secondary_init(unsigned int cpu); void platform_request_ipi(int irq, /*irq_handler_t*/ void *handler); -void platform_send_ipi(cpumask_t callmap, int irq); - void platform_send_ipi_cpu(unsigned int cpu, int irq); void platform_clear_ipi(unsigned int cpu, int irq); diff --git a/arch/blackfin/mach-bf561/smp.c b/arch/blackfin/mach-bf561/smp.c index 7b07740..630e2c2 100644 --- a/arch/blackfin/mach-bf561/smp.c +++ b/arch/blackfin/mach-bf561/smp.c @@ -116,19 +116,6 @@ void __init platform_request_ipi(int irq, void *handler) panic("Cannot request %s for IPI service", name); } -void platform_send_ipi(cpumask_t callmap, int irq) -{ - unsigned int cpu; - int offset = (irq == IRQ_SUPPLE_0) ? 6 : 8; - - for_each_cpu_mask(cpu, callmap) { - BUG_ON(cpu >= 2); - SSYNC(); - bfin_write_SICB_SYSCR(bfin_read_SICB_SYSCR() | (1 << (offset + cpu))); - SSYNC(); - } -} - void platform_send_ipi_cpu(unsigned int cpu, int irq) { int offset = (irq == IRQ_SUPPLE_0) ? 6 : 8; -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly 2011-04-26 1:53 ` (resend) [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro @ 2011-04-26 1:55 ` KOSAKI Motohiro 2011-05-13 2:45 ` Mike Frysinger 2011-04-26 1:56 ` (resend) [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro ` (2 subsequent siblings) 3 siblings, 1 reply; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-26 1:55 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, uclinux-dist-devel, Mike Frysinger, LKML, Andrew Morton We plan to remove cpu_possible_map and cpu_present_map later and we have proper init_cpu_possible() and init_cpu_present() APIs. Therefore this patch rewrites platform_init_cpus and platform_prepare_cpus by their APIs. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Mike Frysinger <vapier@gentoo.org> Cc: uclinux-dist-devel@blackfin.uclinux.org --- arch/blackfin/mach-bf561/smp.c | 14 ++++++++++---- 1 files changed, 10 insertions(+), 4 deletions(-) I sent this parch to wrong address (device-driver-devel@blackfin.uclinux.org). thus now I'm resending it. diff --git a/arch/blackfin/mach-bf561/smp.c b/arch/blackfin/mach-bf561/smp.c index 630e2c2..2fca469 100644 --- a/arch/blackfin/mach-bf561/smp.c +++ b/arch/blackfin/mach-bf561/smp.c @@ -24,17 +24,23 @@ static DEFINE_SPINLOCK(boot_lock); void __init platform_init_cpus(void) { - cpu_set(0, cpu_possible_map); /* CoreA */ - cpu_set(1, cpu_possible_map); /* CoreB */ + struct cpumask mask; + + cpumask_set_cpu(0, &mask); /* CoreA */ + cpumask_set_cpu(1, &mask); /* CoreB */ + init_cpu_possible(&mask); } void __init platform_prepare_cpus(unsigned int max_cpus) { + struct cpumask mask; + bfin_relocate_coreb_l1_mem(); /* Both cores ought to be present on a bf561! */ - cpu_set(0, cpu_present_map); /* CoreA */ - cpu_set(1, cpu_present_map); /* CoreB */ + cpumask_set_cpu(0, &mask); /* CoreA */ + cpumask_set_cpu(1, &mask); /* CoreB */ + init_cpu_present(&mask); } int __init setup_profiling_timer(unsigned int multiplier) /* not supported */ -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly 2011-04-26 1:55 ` (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro @ 2011-05-13 2:45 ` Mike Frysinger 0 siblings, 0 replies; 17+ messages in thread From: Mike Frysinger @ 2011-05-13 2:45 UTC (permalink / raw) To: KOSAKI Motohiro; +Cc: uclinux-dist-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 21:55, KOSAKI Motohiro wrote: > We plan to remove cpu_possible_map and cpu_present_map later and we > have proper init_cpu_possible() and init_cpu_present() APIs. > > Therefore this patch rewrites platform_init_cpus and platform_prepare_cpus > by their APIs. ive merged this patch now, thanks -mike ^ permalink raw reply [flat|nested] 17+ messages in thread
* (resend) [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly 2011-04-26 1:53 ` (resend) [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro 2011-04-26 1:55 ` (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro @ 2011-04-26 1:56 ` KOSAKI Motohiro 2011-05-13 2:47 ` Mike Frysinger 2011-04-26 1:57 ` (resend) [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro 2011-04-26 4:47 ` (resend) [PATCH 1/4] blackfin: remove unused function Mike Frysinger 3 siblings, 1 reply; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-26 1:56 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, uclinux-dist-devel, Mike Frysinger, LKML, Andrew Morton Every callter (except kthread_bind) should use proper set_cpus_allowed_ptr() APIs. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Mike Frysinger <vapier@gentoo.org> Cc: uclinux-dist-devel@blackfin.uclinux.org --- I'm curious why this mysterious code is necessary. Why sys_clone() restrict allowed cpus automatically and why don't it restore the restriction when do_fork() is finished. ps: I sent this parch to wrong address (device-driver-devel@blackfin.uclinux.org). thus now I'm resending it. arch/blackfin/kernel/process.c | 6 ++---- 1 files changed, 2 insertions(+), 4 deletions(-) diff --git a/arch/blackfin/kernel/process.c b/arch/blackfin/kernel/process.c index b407bc8..6a660fa 100644 --- a/arch/blackfin/kernel/process.c +++ b/arch/blackfin/kernel/process.c @@ -171,10 +171,8 @@ asmlinkage int bfin_clone(struct pt_regs *regs) unsigned long newsp; #ifdef __ARCH_SYNC_CORE_DCACHE - if (current->rt.nr_cpus_allowed == num_possible_cpus()) { - current->cpus_allowed = cpumask_of_cpu(smp_processor_id()); - current->rt.nr_cpus_allowed = 1; - } + if (current->rt.nr_cpus_allowed == num_possible_cpus()) + set_cpus_allowed_ptr(current, cpumask_of(smp_processor_id())); #endif /* syscall2 puts clone_flags in r0 and usp in r1 */ -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: (resend) [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly 2011-04-26 1:56 ` (resend) [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro @ 2011-05-13 2:47 ` Mike Frysinger 0 siblings, 0 replies; 17+ messages in thread From: Mike Frysinger @ 2011-05-13 2:47 UTC (permalink / raw) To: KOSAKI Motohiro; +Cc: uclinux-dist-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 21:56, KOSAKI Motohiro wrote: > Every callter (except kthread_bind) should use proper > set_cpus_allowed_ptr() APIs. ive merged this, thanks -mike ^ permalink raw reply [flat|nested] 17+ messages in thread
* (resend) [PATCH 4/4] blackfin: convert old cpumask API to new one 2011-04-26 1:53 ` (resend) [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro 2011-04-26 1:55 ` (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro 2011-04-26 1:56 ` (resend) [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro @ 2011-04-26 1:57 ` KOSAKI Motohiro 2011-05-13 2:51 ` Mike Frysinger 2011-04-26 4:47 ` (resend) [PATCH 1/4] blackfin: remove unused function Mike Frysinger 3 siblings, 1 reply; 17+ messages in thread From: KOSAKI Motohiro @ 2011-04-26 1:57 UTC (permalink / raw) To: KOSAKI Motohiro Cc: kosaki.motohiro, uclinux-dist-devel, Mike Frysinger, LKML, Andrew Morton old cpu_xxx() APIs is planned to removed later. then, converted. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Mike Frysinger <vapier@gentoo.org> Cc: uclinux-dist-devel@blackfin.uclinux.org --- I sent this parch to wrong address (device-driver-devel@blackfin.uclinux.org). thus now I'm resending it. arch/blackfin/kernel/nmi.c | 8 ++++---- arch/blackfin/kernel/setup.c | 4 ++-- arch/blackfin/mach-common/dpmc.c | 7 ++++--- arch/blackfin/mach-common/smp.c | 27 ++++++++++++++------------- 4 files changed, 24 insertions(+), 22 deletions(-) diff --git a/arch/blackfin/kernel/nmi.c b/arch/blackfin/kernel/nmi.c index 0b5f72f..7af8b70 100644 --- a/arch/blackfin/kernel/nmi.c +++ b/arch/blackfin/kernel/nmi.c @@ -145,16 +145,16 @@ int check_nmi_wdt_touched(void) { unsigned int this_cpu = smp_processor_id(); unsigned int cpu; + cpumask_t mask; - cpumask_t mask = cpu_online_map; - + cpumask_copy(&mask, cpu_online_mask); if (!atomic_read(&nmi_touched[this_cpu])) return 0; atomic_set(&nmi_touched[this_cpu], 0); - cpu_clear(this_cpu, mask); - for_each_cpu_mask(cpu, mask) { + cpumask_clear_cpu(this_cpu, &mask); + for_each_cpu(cpu,&mask) { invalidate_dcache_range((unsigned long)(&nmi_touched[cpu]), (unsigned long)(&nmi_touched[cpu])); if (!atomic_read(&nmi_touched[cpu])) diff --git a/arch/blackfin/kernel/setup.c b/arch/blackfin/kernel/setup.c index 805c613..c00c87b 100644 --- a/arch/blackfin/kernel/setup.c +++ b/arch/blackfin/kernel/setup.c @@ -1326,7 +1326,7 @@ static int show_cpuinfo(struct seq_file *m, void *v) static void *c_start(struct seq_file *m, loff_t *pos) { if (*pos == 0) - *pos = first_cpu(cpu_online_map); + *pos = cpumask_first(cpu_online_mask); if (*pos >= num_online_cpus()) return NULL; @@ -1335,7 +1335,7 @@ static void *c_start(struct seq_file *m, loff_t *pos) static void *c_next(struct seq_file *m, void *v, loff_t *pos) { - *pos = next_cpu(*pos, cpu_online_map); + *pos = cpumask_next(*pos, cpu_online_mask); return c_start(m, pos); } diff --git a/arch/blackfin/mach-common/dpmc.c b/arch/blackfin/mach-common/dpmc.c index 382099f..6ea2921 100644 --- a/arch/blackfin/mach-common/dpmc.c +++ b/arch/blackfin/mach-common/dpmc.c @@ -88,10 +88,11 @@ static void bfin_wakeup_cpu(void) { unsigned int cpu; unsigned int this_cpu = smp_processor_id(); - cpumask_t mask = cpu_online_map; + cpumask_t mask; - cpu_clear(this_cpu, mask); - for_each_cpu_mask(cpu, mask) + cpumask_copy(&mask, cpu_online_mask); + cpumask_clear_cpu(this_cpu, &mask); + for_each_cpu(cpu, &mask) platform_send_ipi_cpu(cpu, IRQ_SUPPLE_0); } diff --git a/arch/blackfin/mach-common/smp.c b/arch/blackfin/mach-common/smp.c index 1fbd94c..f430bba 100644 --- a/arch/blackfin/mach-common/smp.c +++ b/arch/blackfin/mach-common/smp.c @@ -96,7 +96,7 @@ static void ipi_cpu_stop(unsigned int cpu) dump_stack(); spin_unlock(&stop_lock); - cpu_clear(cpu, cpu_online_map); + set_cpu_online(cpu, false); local_irq_disable(); @@ -146,7 +146,7 @@ static void ipi_call_function(unsigned int cpu, struct ipi_message *msg) */ resync_core_dcache(); #endif - cpu_clear(cpu, *msg->call_struct.waitmask); + cpumask_clear_cpu(cpu, msg->call_struct.waitmask); } } @@ -222,9 +222,10 @@ static inline void smp_send_message(cpumask_t callmap, unsigned long type, struct ipi_message_queue *msg_queue; struct ipi_message *msg; unsigned long flags, next_msg; - cpumask_t waitmask = callmap; /* waitmask is shared by all cpus */ + cpumask_t waitmask; /* waitmask is shared by all cpus */ - for_each_cpu_mask(cpu, callmap) { + cpumask_copy(&waitmask, &callmap); + for_each_cpu(cpu, &callmap) { msg_queue = &per_cpu(ipi_msg_queue, cpu); spin_lock_irqsave(&msg_queue->lock, flags); if (msg_queue->count < BFIN_IPI_MSGQ_LEN) { @@ -246,7 +247,7 @@ static inline void smp_send_message(cpumask_t callmap, unsigned long type, } if (wait) { - while (!cpus_empty(waitmask)) + while (!cpumask_empty(&waitmask)) blackfin_dcache_invalidate_range( (unsigned long)(&waitmask), (unsigned long)(&waitmask)); @@ -265,9 +266,9 @@ int smp_call_function(void (*func)(void *info), void *info, int wait) cpumask_t callmap; preempt_disable(); - callmap = cpu_online_map; - cpu_clear(smp_processor_id(), callmap); - if (!cpus_empty(callmap)) + cpumask_copy(&callmap, cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), &callmap); + if (!cpumask_empty(&callmap)) smp_send_message(callmap, BFIN_IPI_CALL_FUNC, func, info, wait); preempt_enable(); @@ -284,8 +285,8 @@ int smp_call_function_single(int cpuid, void (*func) (void *info), void *info, if (cpu_is_offline(cpu)) return 0; - cpus_clear(callmap); - cpu_set(cpu, callmap); + cpumask_clear(&callmap); + cpumask_set_cpu(cpu, &callmap); smp_send_message(callmap, BFIN_IPI_CALL_FUNC, func, info, wait); @@ -308,9 +309,9 @@ void smp_send_stop(void) cpumask_t callmap; preempt_disable(); - callmap = cpu_online_map; - cpu_clear(smp_processor_id(), callmap); - if (!cpus_empty(callmap)) + cpumask_copy(&callmap, cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), &callmap); + if (!cpumask_empty(&callmap)) smp_send_message(callmap, BFIN_IPI_CPU_STOP, NULL, NULL, 0); preempt_enable(); -- 1.7.3.1 ^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: (resend) [PATCH 4/4] blackfin: convert old cpumask API to new one 2011-04-26 1:57 ` (resend) [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro @ 2011-05-13 2:51 ` Mike Frysinger 0 siblings, 0 replies; 17+ messages in thread From: Mike Frysinger @ 2011-05-13 2:51 UTC (permalink / raw) To: KOSAKI Motohiro; +Cc: uclinux-dist-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 21:57, KOSAKI Motohiro wrote: > old cpu_xxx() APIs is planned to removed later. then, converted. ive merged this one too, thanks -mike ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: (resend) [PATCH 1/4] blackfin: remove unused function 2011-04-26 1:53 ` (resend) [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro ` (2 preceding siblings ...) 2011-04-26 1:57 ` (resend) [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro @ 2011-04-26 4:47 ` Mike Frysinger 3 siblings, 0 replies; 17+ messages in thread From: Mike Frysinger @ 2011-04-26 4:47 UTC (permalink / raw) To: KOSAKI Motohiro; +Cc: uclinux-dist-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 21:53, KOSAKI Motohiro wrote: > platform_send_ipi() has old call-by-value cpumask_t fashon and > it's unused. looking at callers of platform_send_ipi_cpu(), it seems that there's at least one place that could (should) be converted to platform_send_ipi(). i'd rather update this func to the new api so i can make that conversion. -mike ^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 0/4] blackfin: convet cpumask apis 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro ` (4 preceding siblings ...) 2011-04-25 18:11 ` [PATCH 0/4] blackfin: convet cpumask apis Mike Frysinger @ 2011-04-26 18:26 ` Thiago Farina 5 siblings, 0 replies; 17+ messages in thread From: Thiago Farina @ 2011-04-26 18:26 UTC (permalink / raw) To: KOSAKI Motohiro Cc: Michael Hennerich, device-driver-devel, LKML, Andrew Morton On Mon, Apr 25, 2011 at 6:27 AM, KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> wrote: > > Rusty Russle introduced a lot of cpumask related APIs. and He gone even s/Russle/Russell s/and He/And he/ > though the work is unfinished. This patch series is sequel of his work. > > No functional change. only api change. > > note: I've confirmed only cross-compile build. so, I hope blackfin > developers see the code. > > > KOSAKI Motohiro (4): > blackfin: remove unused function > blackfin: don't touch cpu_possible_map and cpu_present_map directly > blackfin: don't touch task->cpus_allowed directly > blackfin: convert old cpumask API to new one > > arch/blackfin/kernel/nmi.c | 8 ++-- > arch/blackfin/kernel/process.c | 6 +-- > arch/blackfin/kernel/setup.c | 4 +- > arch/blackfin/mach-bf561/include/mach/smp.h | 2 - > arch/blackfin/mach-bf561/smp.c | 27 ++++------- > arch/blackfin/mach-common/dpmc.c | 7 ++- > arch/blackfin/mach-common/smp.c | 27 ++++++----- > > > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ > ^ permalink raw reply [flat|nested] 17+ messages in thread
end of thread, other threads:[~2011-05-13 2:51 UTC | newest] Thread overview: 17+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2011-04-25 9:27 [PATCH 0/4] blackfin: convet cpumask apis KOSAKI Motohiro 2011-04-25 9:28 ` [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro 2011-04-25 9:29 ` [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro 2011-04-25 9:31 ` [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro 2011-04-25 9:32 ` [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro 2011-04-25 18:11 ` [PATCH 0/4] blackfin: convet cpumask apis Mike Frysinger 2011-04-25 18:18 ` Mike Frysinger 2011-04-26 1:49 ` KOSAKI Motohiro 2011-04-26 1:53 ` (resend) [PATCH 1/4] blackfin: remove unused function KOSAKI Motohiro 2011-04-26 1:55 ` (resend) [PATCH 2/4] blackfin: don't touch cpu_possible_map and cpu_present_map directly KOSAKI Motohiro 2011-05-13 2:45 ` Mike Frysinger 2011-04-26 1:56 ` (resend) [PATCH 3/4] blackfin: don't touch task->cpus_allowed directly KOSAKI Motohiro 2011-05-13 2:47 ` Mike Frysinger 2011-04-26 1:57 ` (resend) [PATCH 4/4] blackfin: convert old cpumask API to new one KOSAKI Motohiro 2011-05-13 2:51 ` Mike Frysinger 2011-04-26 4:47 ` (resend) [PATCH 1/4] blackfin: remove unused function Mike Frysinger 2011-04-26 18:26 ` [PATCH 0/4] blackfin: convet cpumask apis Thiago Farina
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).