lkml.org 
[lkml]   [2008]   [Sep]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[RFC 13/13] genapic: reduce stack pressuge in io_apic.c step 5 cpu_mask_to_apicid
  * Step 5 "cpu_mask_to_apicid" of cleaning up io_apic.c modifies the
cpu_mask_to_apicid genapic interface to pass a pointer to incoming
cpumask_t argument removing yet another "cpumask_t variable on the stack".

apicid = cpu_mask_to_apicid(cpumask);

becomes:

apicid = cpu_mask_to_apicid(&cpumask);

* All the appropriate genapic "cpu_mask_to_apicid" functions
are modified to use this new interface.

* Total stack savings are:

====== Stack (-l 100)

1 - initial
2 - stack-hogs-io_apic_c-step-1-temp-cpumask_ts
3 - stack-hogs-io_apic_c-step-2-internal-abi
4 - stack-hogs-io_apic_c-step-3-target_cpus
5 - stack-hogs-io_apic_c-step-4-vector_allocation_domain
6 - stack-hogs-io_apic_c-step-5-cpu_mask_to_apicid

.1. .2. .3. .4. .5. .6. ..final..
1576 -512 . -512 . -552 . -100% setup_IO_APIC_irq
1560 -512 . -528 . -520 . -100% msi_compose_msg
1560 -512 . -512 . -536 . -100% arch_setup_ht_irq
1560 -1008 . . -552 . . -100% __assign_irq_vector
1144 . -512 -632 . . . -100% setup_IO_APIC
1048 . . -512 . -536 . -100% setup_timer_IRQ0_pin
1048 . -512 -536 . . . -100% setup_ioapic_dest
1048 -512 -536 . . . . -100% set_ioapic_affinity_irq
1040 -512 . . . -528 . -100% set_msi_irq_affinity
1040 -512 . . . -528 . -100% set_ht_irq_affinity
1040 -512 . . . -528 . -100% dmar_msi_set_affinity
1032 -8 -512 -512 . . . -100% create_irq_nr
1024 . . . -1024 . . -100% x2apic_vector_allocation_domain
1024 . . . -1024 . . -100% uv_vector_allocation_domain
520 -520 . . . . . -100% __clear_irq_vector
512 -512 . . . . . -100% irq_complete_move
512 . -512 . . . . -100% assign_irq_vector
0 . +536 . . -536 . . set_ioapic_affinity_irq_p
18288 -5632 -2048 -3744 -2600 -4264 . -100% Totals

Applies to linux-2.6.tip/master.

Signed-off-by: Mike Travis <travis@sgi.com>
---
arch/x86/kernel/genapic_flat_64.c | 8 ++++----
arch/x86/kernel/genx2apic_cluster.c | 4 ++--
arch/x86/kernel/genx2apic_phys.c | 4 ++--
arch/x86/kernel/genx2apic_uv_x.c | 4 ++--
arch/x86/kernel/io_apic.c | 20 ++++++++++----------
include/asm-x86/bigsmp/apic.h | 4 ++--
include/asm-x86/es7000/apic.h | 8 ++++----
include/asm-x86/genapic_32.h | 2 +-
include/asm-x86/genapic_64.h | 2 +-
include/asm-x86/mach-default/mach_apic.h | 4 ++--
include/asm-x86/numaq/apic.h | 2 +-
include/asm-x86/summit/apic.h | 8 ++++----
12 files changed, 35 insertions(+), 35 deletions(-)

--- linux-2.6.tip.orig/arch/x86/kernel/genapic_flat_64.c
+++ linux-2.6.tip/arch/x86/kernel/genapic_flat_64.c
@@ -140,9 +140,9 @@ static int flat_apic_id_registered(void)
return physid_isset(read_xapic_id(), phys_cpu_present_map);
}

-static unsigned int flat_cpu_mask_to_apicid(cpumask_t cpumask)
+static unsigned int flat_cpu_mask_to_apicid(const cpumask_t *cpumask)
{
- return cpus_addr(cpumask)[0] & APIC_ALL_CPUS;
+ return cpus_addr(*cpumask)[0] & APIC_ALL_CPUS;
}

static unsigned int phys_pkg_id(int index_msb)
@@ -227,7 +227,7 @@ static void physflat_send_IPI_all(int ve
physflat_send_IPI_mask(&cpu_online_map, vector);
}

-static unsigned int physflat_cpu_mask_to_apicid(cpumask_t cpumask)
+static unsigned int physflat_cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int cpu;

@@ -235,7 +235,7 @@ static unsigned int physflat_cpu_mask_to
* We're using fixed IRQ delivery, can only return one phys APIC ID.
* May as well be the first.
*/
- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
if ((unsigned)cpu < nr_cpu_ids)
return per_cpu(x86_cpu_to_apicid, cpu);
else
--- linux-2.6.tip.orig/arch/x86/kernel/genx2apic_cluster.c
+++ linux-2.6.tip/arch/x86/kernel/genx2apic_cluster.c
@@ -92,7 +92,7 @@ static int x2apic_apic_id_registered(voi
return 1;
}

-static unsigned int x2apic_cpu_mask_to_apicid(cpumask_t cpumask)
+static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int cpu;

@@ -100,7 +100,7 @@ static unsigned int x2apic_cpu_mask_to_a
* We're using fixed IRQ delivery, can only return one phys APIC ID.
* May as well be the first.
*/
- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
if ((unsigned)cpu < NR_CPUS)
return per_cpu(x86_cpu_to_logical_apicid, cpu);
else
--- linux-2.6.tip.orig/arch/x86/kernel/genx2apic_phys.c
+++ linux-2.6.tip/arch/x86/kernel/genx2apic_phys.c
@@ -85,7 +85,7 @@ static int x2apic_apic_id_registered(voi
return 1;
}

-static unsigned int x2apic_cpu_mask_to_apicid(cpumask_t cpumask)
+static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int cpu;

@@ -93,7 +93,7 @@ static unsigned int x2apic_cpu_mask_to_a
* We're using fixed IRQ delivery, can only return one phys APIC ID.
* May as well be the first.
*/
- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
if ((unsigned)cpu < NR_CPUS)
return per_cpu(x86_cpu_to_apicid, cpu);
else
--- linux-2.6.tip.orig/arch/x86/kernel/genx2apic_uv_x.c
+++ linux-2.6.tip/arch/x86/kernel/genx2apic_uv_x.c
@@ -160,7 +160,7 @@ static void uv_init_apic_ldr(void)
{
}

-static unsigned int uv_cpu_mask_to_apicid(cpumask_t cpumask)
+static unsigned int uv_cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int cpu;

@@ -168,7 +168,7 @@ static unsigned int uv_cpu_mask_to_apici
* We're using fixed IRQ delivery, can only return one phys APIC ID.
* May as well be the first.
*/
- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
if ((unsigned)cpu < nr_cpu_ids)
return per_cpu(x86_cpu_to_apicid, cpu);
else
--- linux-2.6.tip.orig/arch/x86/kernel/io_apic.c
+++ linux-2.6.tip/arch/x86/kernel/io_apic.c
@@ -602,7 +602,7 @@ static void set_ioapic_affinity_irq_p(un
}

cpus_and(*tmp, cfg->domain, *mask);
- dest = cpu_mask_to_apicid(*tmp);
+ dest = cpu_mask_to_apicid(tmp);
/*
* Only the high 8 bits are valid.
*/
@@ -1546,7 +1546,7 @@ static void setup_IO_APIC_irq(int apic,


if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry,
- cpu_mask_to_apicid(*mask), trigger, polarity,
+ cpu_mask_to_apicid(mask), trigger, polarity,
cfg->vector)) {
printk("Failed to setup ioapic entry for ioapic %d, pin %d\n",
mp_ioapics[apic].mp_apicid, pin);
@@ -1626,7 +1626,7 @@ static void __init setup_timer_IRQ0_pin(
*/
entry.dest_mode = INT_DEST_MODE;
entry.mask = 1; /* mask IRQ now */
- entry.dest = cpu_mask_to_apicid(*tgt_cpus);
+ entry.dest = cpu_mask_to_apicid(tgt_cpus);
entry.delivery_mode = INT_DELIVERY_MODE;
entry.polarity = 0;
entry.trigger = 0;
@@ -2322,7 +2322,7 @@ static void migrate_ioapic_irq(int irq,

cfg = irq_cfg(irq);
cpus_and(*tmpmask, cfg->domain, *mask);
- dest = cpu_mask_to_apicid(*tmpmask);
+ dest = cpu_mask_to_apicid(tmpmask);

desc = irq_to_desc(irq);
modify_ioapic_rte = desc->status & IRQ_LEVEL;
@@ -3204,7 +3204,7 @@ static int msi_compose_msg(struct pci_de

cfg = irq_cfg(irq);
cpus_and(*tgt_cpus, cfg->domain, *tgt_cpus);
- dest = cpu_mask_to_apicid(*tgt_cpus);
+ dest = cpu_mask_to_apicid(tgt_cpus);
put_cpumask_var(tgt_cpus, cpumask_irq_level_4);

#ifdef CONFIG_INTR_REMAP
@@ -3277,7 +3277,7 @@ static void set_msi_irq_affinity(unsigne

cfg = irq_cfg(irq);
cpus_and(*tmp, cfg->domain, mask);
- dest = cpu_mask_to_apicid(*tmp);
+ dest = cpu_mask_to_apicid(tmp);

read_msi_msg(irq, &msg);

@@ -3319,7 +3319,7 @@ static void ir_set_msi_irq_affinity(unsi

cfg = irq_cfg(irq);
cpus_and(*tmp, cfg->domain, mask);
- dest = cpu_mask_to_apicid(*tmp);
+ dest = cpu_mask_to_apicid(tmp);

irte.vector = cfg->vector;
irte.dest_id = IRTE_DEST(dest);
@@ -3561,7 +3561,7 @@ static void dmar_msi_set_affinity(unsign

cfg = irq_cfg(irq);
cpus_and(*tmp, cfg->domain, mask);
- dest = cpu_mask_to_apicid(*tmp);
+ dest = cpu_mask_to_apicid(tmp);

dmar_msi_read(irq, &msg);

@@ -3643,7 +3643,7 @@ static void set_ht_irq_affinity(unsigned

cfg = irq_cfg(irq);
cpus_and(*tmp, cfg->domain, mask);
- dest = cpu_mask_to_apicid(*tmp);
+ dest = cpu_mask_to_apicid(tmp);

target_ht_irq(irq, dest, cfg->vector);
desc = irq_to_desc(irq);
@@ -3679,7 +3679,7 @@ int arch_setup_ht_irq(unsigned int irq,

cfg = irq_cfg(irq);
cpus_and(*tgt_cpus, cfg->domain, *tgt_cpus);
- dest = cpu_mask_to_apicid(*tgt_cpus);
+ dest = cpu_mask_to_apicid(tgt_cpus);

msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest);

--- linux-2.6.tip.orig/include/asm-x86/bigsmp/apic.h
+++ linux-2.6.tip/include/asm-x86/bigsmp/apic.h
@@ -121,12 +121,12 @@ static inline int check_phys_apicid_pres
}

/* As we are using single CPU as destination, pick only one CPU here */
-static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
+static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int cpu;
int apicid;

- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
apicid = cpu_to_logical_apicid(cpu);
return apicid;
}
--- linux-2.6.tip.orig/include/asm-x86/es7000/apic.h
+++ linux-2.6.tip/include/asm-x86/es7000/apic.h
@@ -144,14 +144,14 @@ static inline int check_phys_apicid_pres
return (1);
}

-static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
+static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int num_bits_set;
int cpus_found = 0;
int cpu;
int apicid;

- num_bits_set = cpus_weight(cpumask);
+ num_bits_set = cpus_weight(*cpumask);
/* Return id to all */
if (num_bits_set == NR_CPUS)
#if defined CONFIG_ES7000_CLUSTERED_APIC
@@ -163,10 +163,10 @@ static inline unsigned int cpu_mask_to_a
* The cpus in the mask must all be on the apic cluster. If are not
* on the same apicid cluster return default value of TARGET_CPUS.
*/
- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
apicid = cpu_to_logical_apicid(cpu);
while (cpus_found < num_bits_set) {
- if (cpu_isset(cpu, cpumask)) {
+ if (cpu_isset(cpu, *cpumask)) {
int new_apicid = cpu_to_logical_apicid(cpu);
if (apicid_cluster(apicid) !=
apicid_cluster(new_apicid)){
--- linux-2.6.tip.orig/include/asm-x86/genapic_32.h
+++ linux-2.6.tip/include/asm-x86/genapic_32.h
@@ -57,7 +57,7 @@ struct genapic {

unsigned (*get_apic_id)(unsigned long x);
unsigned long apic_id_mask;
- unsigned int (*cpu_mask_to_apicid)(cpumask_t cpumask);
+ unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask);
void (*vector_allocation_domain)(int cpu, cpumask_t *retmask);

#ifdef CONFIG_SMP
--- linux-2.6.tip.orig/include/asm-x86/genapic_64.h
+++ linux-2.6.tip/include/asm-x86/genapic_64.h
@@ -29,7 +29,7 @@ struct genapic {
void (*send_IPI_all)(int vector);
void (*send_IPI_self)(int vector);
/* */
- unsigned int (*cpu_mask_to_apicid)(cpumask_t cpumask);
+ unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask);
unsigned int (*phys_pkg_id)(int index_msb);
unsigned int (*get_apic_id)(unsigned long x);
unsigned long (*set_apic_id)(unsigned int id);
--- linux-2.6.tip.orig/include/asm-x86/mach-default/mach_apic.h
+++ linux-2.6.tip/include/asm-x86/mach-default/mach_apic.h
@@ -65,9 +65,9 @@ static inline int apic_id_registered(voi
return physid_isset(read_apic_id(), phys_cpu_present_map);
}

-static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
+static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask)
{
- return cpus_addr(cpumask)[0];
+ return cpus_addr(*cpumask)[0];
}

static inline u32 phys_pkg_id(u32 cpuid_apic, int index_msb)
--- linux-2.6.tip.orig/include/asm-x86/numaq/apic.h
+++ linux-2.6.tip/include/asm-x86/numaq/apic.h
@@ -122,7 +122,7 @@ static inline void enable_apic_mode(void
* We use physical apicids here, not logical, so just return the default
* physical broadcast to stop people from breaking us
*/
-static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
+static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask)
{
return (int) 0xF;
}
--- linux-2.6.tip.orig/include/asm-x86/summit/apic.h
+++ linux-2.6.tip/include/asm-x86/summit/apic.h
@@ -137,14 +137,14 @@ static inline void enable_apic_mode(void
{
}

-static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask)
+static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask)
{
int num_bits_set;
int cpus_found = 0;
int cpu;
int apicid;

- num_bits_set = cpus_weight(cpumask);
+ num_bits_set = cpus_weight(*cpumask);
/* Return id to all */
if (num_bits_set == NR_CPUS)
return (int) 0xFF;
@@ -152,10 +152,10 @@ static inline unsigned int cpu_mask_to_a
* The cpus in the mask must all be on the apic cluster. If are not
* on the same apicid cluster return default value of TARGET_CPUS.
*/
- cpu = first_cpu(cpumask);
+ cpu = first_cpu(*cpumask);
apicid = cpu_to_logical_apicid(cpu);
while (cpus_found < num_bits_set) {
- if (cpu_isset(cpu, cpumask)) {
+ if (cpu_isset(cpu, *cpumask)) {
int new_apicid = cpu_to_logical_apicid(cpu);
if (apicid_cluster(apicid) !=
apicid_cluster(new_apicid)){
--


\
 
 \ /
  Last update: 2008-09-07 01:57    [W:0.165 / U:0.492 seconds]
©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site