powerpc: Abstract indexing of lppaca structs
PATCH 1/2 powerpc: Abstract indexing of lppaca structs
Currently we have the lppaca structs as a simple array of NR_CPUS entries, taking up space in the data section of the kernel image. In future we would like to allocate them dynamically, so this abstracts out the accesses to the array, making it easier to change how we locate the lppaca for a given cpu in future. Specifically, lppaca[cpu] changes to lppaca_of(cpu).
Signed-off-by Paul Mackerras <[email protected]>
--- arch/powerpc/include/asm/lppaca.h | 2 ++ arch/powerpc/kernel/lparcfg.c | 14 +++++++------- arch/powerpc/lib/locks.c | 4 ++-- arch/powerpc/platforms/iseries/dt.c | 4 ++-- arch/powerpc/platforms/iseries/smp.c | 2 +- arch/powerpc/platforms/pseries/dtl.c | 8 ++++---- arch/powerpc/platforms/pseries/lpar.c | 4 ++-- 7 files changed, 20 insertions(+), 18 deletions(-)
arch/powerpc/include/asm/lppaca.h +2 -0
@@ -153,6 +153,8 @@ struct lppaca {extern struct lppaca lppaca[]; +#define lppaca_of (cpu) (lppaca[cpu]) +/* * SLB shadow buffer structure as defined in the PAPR. The save_area * contains adjacent ESID and VSID pairs for each shadowed SLB. The
arch/powerpc/kernel/lparcfg.c +7 -7
@@ -56,7 +56,7 @@ static unsigned long get_purr(void)for_each_possible_cpu (cpu) {if (firmware_has_feature (FW_FEATURE_ISERIES )) - sum_purr += lppaca[cpu].emulated_time_base; + sum_purr +=lppaca_of (cpu).emulated_time_base;else {struct cpu_usage *cu;
@@ -263,7 +263,7 @@ static void parse_ppp_data(struct seq_file *m) ppp_data.active_system_procs);/* pool related entries are apropriate for shared configs */ -if (lppaca[0 ].shared_proc) { +if (lppaca_of (0 ).shared_proc) {unsigned long pool_idle_time, pool_procs;seq_printf (m,"pool=%d\n" , ppp_data.pool_num);
@@ -460,8 +460,8 @@ static void pseries_cmo_data(struct seq_file *m)return ;for_each_possible_cpu (cpu) { - cmo_faults += lppaca[cpu].cmo_faults; - cmo_fault_time += lppaca[cpu].cmo_fault_time; + cmo_faults +=lppaca_of (cpu).cmo_faults; + cmo_fault_time +=lppaca_of (cpu).cmo_fault_time; }seq_printf (m,"cmo_faults=%lu\n" , cmo_faults);
@@ -479,8 +479,8 @@ static void splpar_dispatch_data(struct seq_file *m)unsigned long dispatch_dispersions =0 ;for_each_possible_cpu (cpu) { - dispatches += lppaca[cpu].yield_count; - dispatch_dispersions += lppaca[cpu].dispersion_count; + dispatches +=lppaca_of (cpu).yield_count; + dispatch_dispersions +=lppaca_of (cpu).dispersion_count; }seq_printf (m,"dispatches=%lu\n" , dispatches);
@@ -545,7 +545,7 @@ static int pseries_lparcfg_data(struct seq_file *m, void *v)seq_printf (m,"partition_potential_processors=%d\n" , partition_potential_processors); -seq_printf (m,"shared_processor_mode=%d\n" , lppaca[0 ].shared_proc); +seq_printf (m,"shared_processor_mode=%d\n" ,lppaca_of (0 ).shared_proc);seq_printf (m,"slb_size=%d\n" , mmu_slb_size);
arch/powerpc/lib/locks.c +2 -2
@@ -34,7 +34,7 @@ void __spin_yield(arch_spinlock_t *lock)return ; holder_cpu = lock_value &0xffff ;BUG_ON (holder_cpu >=NR_CPUS ); - yield_count = lppaca[holder_cpu].yield_count; + yield_count =lppaca_of (holder_cpu).yield_count;if ((yield_count &1 ) ==0 )return ;/* virtual cpu is currently running */ rmb ();
@@ -65,7 +65,7 @@ void __rw_yield(arch_rwlock_t *rw)return ;/* no write lock at present */ holder_cpu = lock_value &0xffff ;BUG_ON (holder_cpu >=NR_CPUS ); - yield_count = lppaca[holder_cpu].yield_count; + yield_count =lppaca_of (holder_cpu).yield_count;if ((yield_count &1 ) ==0 )return ;/* virtual cpu is currently running */ rmb ();
arch/powerpc/platforms/iseries/dt.c +2 -2
@@ -243,7 +243,7 @@ static void __init dt_cpus(struct iseries_flat_dt *dt) pft_size[1 ] =__ilog2 (HvCallHpt_getHptPages () *HW_PAGE_SIZE );for (i =0 ; i <NR_CPUS ; i++) { -if (lppaca[i].dyn_proc_status >=2 ) +if (lppaca_of (i).dyn_proc_status >=2 )continue ;snprintf (p,32 - (p - buf),"@%d" , i);
@@ -251,7 +251,7 @@ static void __init dt_cpus(struct iseries_flat_dt *dt)dt_prop_str (dt,"device_type" , device_type_cpu); - index = lppaca[i].dyn_hv_phys_proc_index; + index =lppaca_of (i).dyn_hv_phys_proc_index; d = &xIoHriProcessorVpd[index];dt_prop_u32 (dt,"i-cache-size" , d->xInstCacheSize *1024 );
arch/powerpc/platforms/iseries/smp.c +1 -1
@@ -91,7 +91,7 @@ static void smp_iSeries_kick_cpu(int nr)BUG_ON ((nr <0 ) || (nr >=NR_CPUS ));/* Verify that our partition has a processor nr */ -if (lppaca[nr].dyn_proc_status >=2 ) +if (lppaca_of (nr).dyn_proc_status >=2 )return ;/* The processor is currently spinning, waiting
arch/powerpc/platforms/pseries/dtl.c +4 -4
@@ -107,14 +107,14 @@ static int dtl_enable(struct dtl *dtl) }/* set our initial buffer indices */ - dtl->last_idx = lppaca[dtl->cpu].dtl_idx =0 ; + dtl->last_idx =lppaca_of (dtl->cpu).dtl_idx =0 ;/* ensure that our updates to the lppaca fields have occurred before * we actually enable the logging */ smp_wmb ();/* enable event logging */ - lppaca[dtl->cpu].dtl_enable_mask = dtl_event_mask; +lppaca_of (dtl->cpu).dtl_enable_mask = dtl_event_mask;return 0 ; }
@@ -123,7 +123,7 @@ static void dtl_disable(struct dtl *dtl) {int hwcpu =get_hard_smp_processor_id (dtl->cpu); - lppaca[dtl->cpu].dtl_enable_mask =0x0 ; +lppaca_of (dtl->cpu).dtl_enable_mask =0x0 ;unregister_dtl (hwcpu,__pa (dtl->buf));
@@ -171,7 +171,7 @@ static ssize_t dtl_file_read(struct file *filp, char __user *buf, size_t len,/* actual number of entries read */ n_read =0 ; - cur_idx = lppaca[dtl->cpu].dtl_idx; + cur_idx =lppaca_of (dtl->cpu).dtl_idx; last_idx = dtl->last_idx;if (cur_idx - last_idx > dtl->buf_entries) {
arch/powerpc/platforms/pseries/lpar.c +2 -2
@@ -250,9 +250,9 @@ void vpa_init(int cpu)long ret;if (cpu_has_feature (CPU_FTR_ALTIVEC )) - lppaca[cpu].vmxregs_in_use =1 ; +lppaca_of (cpu).vmxregs_in_use =1 ; - addr =__pa (&lppaca[cpu]); + addr =__pa (&lppaca_of (cpu)); ret =register_vpa (hwcpu, addr);if (ret) {
Signature · 1 line
1.7.1
PATCH 2/2 powerpc: Dynamically allocate most lppaca structs
This arranges for the lppaca structs for most cpus to be dynamically allocated in the same manner as the paca structs. If we don't include support for legacy iSeries, only the first lppaca is statically allocated; the rest are dynamically allocated. If we include legacy iSeries support, then we statically allocate the first 64 lppaca structs, since the iSeries hypervisor requires that the lppaca structs be present in the data section of the kernel image, but legacy iSeries supports at most 64 cpus.
With CONFIG_NR_CPUS, the kernel image size for a typical pSeries config went from:
text data bss dec hex filename
9524478 4734564 8469944 22728986 15ad11a ../test-1024/vmlinux
to:
text data bss dec hex filename
9524482 3751508 8469944 21745934 14bd10e ../test-1024/vmlinux
a reduction of 983052 bytes overall.
Signed-off-by Paul Mackerras <[email protected]>
--- arch/powerpc/include/asm/lppaca.h | 2 +- arch/powerpc/kernel/paca.c | 70 +++++++++++++++++++++++++++++++++++- 2 files changed, 69 insertions(+), 3 deletions(-)
arch/powerpc/include/asm/lppaca.h +1 -1
@@ -153,7 +153,7 @@ struct lppaca {extern struct lppaca lppaca[]; -#define lppaca_of (cpu) (lppaca[cpu]) +#define lppaca_of (cpu) (*paca[cpu].lppaca_ptr)/* * SLB shadow buffer structure as defined in the PAPR. The save_area
arch/powerpc/kernel/paca.c +68 -2
104 lines not on this page · raw message
Signature · 1 line
1.7.1