x86: allocate cpumask during check irq vectors
PATCH 1/1 x86: allocate cpumask during check irq vectors
Fix warning:arch/x86/kernel/irq.c: In function check_irq_vectors_for_cpu_disable:arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052 bytes is larger than 2048 bytes
when NR_CPUS=8192
We should use zalloc_cpumask_var() instead.
-v2: update to GFP_ATOMIC instead and free the allocated cpumask at last.
--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)
arch/x86/kernel/irq.c +17 -7
@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; + +if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + } this_cpu =smp_processor_id (); -cpumask_copy (&online_new, cpu_online_mask); -cpu_clear (this_cpu, online_new); +cpumask_copy (online_new, cpu_online_mask); +cpumask_clear_cpu (this_cpu, online_new); this_count =0 ;for (vector =FIRST_EXTERNAL_VECTOR ; vector <NR_VECTORS ; vector++) {
@@ -289,8 +296,8 @@ int check_irq_vectors_for_cpu_disable(voif (irq >=0 ) { desc =irq_to_desc (irq); data =irq_desc_get_irq_data (desc); -cpumask_copy (&affinity_new, data->affinity); -cpu_clear (this_cpu, affinity_new); +cpumask_copy (affinity_new, data->affinity); +cpumask_clear_cpu (this_cpu, affinity_new);/* Do not count inactive or per-cpu irqs. */ if (!irq_has_action (irq) ||irqd_is_per_cpu (data))
@@ -311,12 +318,15 @@ int check_irq_vectors_for_cpu_disable(vo* mask is not zero; that is the down'd cpu is the * last online cpu in a user set affinity mask. */ -if (cpumask_empty (&affinity_new) || - !cpumask_subset (&affinity_new, &online_new)) +if (cpumask_empty (affinity_new) || + !cpumask_subset (affinity_new, online_new)) this_count++; } } +free_cpumask_var (affinity_new); +free_cpumask_var (online_new); + count =0 ;for_each_online_cpu (cpu) {if (cpu == this_cpu)
Quoted 24 lines · Yinghai Lu
Fix warning:
arch/x86/kernel/irq.c: In functioncheck_irq_vectors_for_cpu_disable:arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052 bytes is larger than 2048 byteswhen NR_CPUS=8192
We should use
zalloc_cpumask_var()instead.-v2: update to
GFP_ATOMICinstead and free the allocated cpumask at last.Signed-off-by Yinghai Lu <[email protected]>Cc Prarit Bhargava <[email protected]>--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)Index:
linux-2.6/arch/x86/kernel/irq.c
===================================================================arch/x86/kernel/irq.c +0
-0@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;
arch/x86/kernel/irq.c int check_irq_vectors_for_cpu_disable(vo · line 287
Quoted 10 lines
arch/x86/kernel/irq.c +8
-1struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; + +if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + }
Atomic allocations can fail easily if the system is under duress.
Thanks,
Ingo
Quoted 31 lines · Ingo Molnar
Quoted 28 lines · Yinghai Lu
Fix warning:
arch/x86/kernel/irq.c: In functioncheck_irq_vectors_for_cpu_disable:arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052 bytes is larger than 2048 byteswhen NR_CPUS=8192
We should use
zalloc_cpumask_var()instead.-v2: update to
GFP_ATOMICinstead and free the allocated cpumask at last.Signed-off-by Yinghai Lu <[email protected]>Cc Prarit Bhargava <[email protected]>--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)Index:
linux-2.6/arch/x86/kernel/irq.c
===================================================================arch/x86/kernel/irq.c +2
-1@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; +
Quoted 8 lines
arch/x86/kernel/irq.c int check_irq_vectors_for_cpu_disable(vo · line 287
Quoted 6 lines
arch/x86/kernel/irq.c +6
-0+if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + }Atomic allocations can fail easily if the system is under duress.
Then the hotplug attempt will fail which IMO is okay. With GFP_KERNEL we might hang the system.
Ingo -- the original submit (and a Reviewed-by: Cheng Gong) is here,
http://marc.info/?l=linux-kernel&m=139035727501266&w=2
Thanks,
P.
Quoted 4 lines
Thanks,
Ingo
Quoted 41 lines · Prarit Bhargava
Quoted 31 lines · Ingo Molnar
Quoted 28 lines · Yinghai Lu
Fix warning:
arch/x86/kernel/irq.c: In functioncheck_irq_vectors_for_cpu_disable:arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052 bytes is larger than 2048 byteswhen NR_CPUS=8192
We should use
zalloc_cpumask_var()instead.-v2: update to
GFP_ATOMICinstead and free the allocated cpumask at last.Signed-off-by Yinghai Lu <[email protected]>Cc Prarit Bhargava <[email protected]>--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)Index:
linux-2.6/arch/x86/kernel/irq.c
===================================================================arch/x86/kernel/irq.c +2
-1@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; +Quoted 8 lines
arch/x86/kernel/irq.c int check_irq_vectors_for_cpu_disable(vo · line 287
Quoted 6 lines
arch/x86/kernel/irq.c +6
-0+if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + }Atomic allocations can fail easily if the system is under duress.
Quoted 1 line
Then the hotplug attempt will fail which IMO is okay. [...]
Which is not OK at all for reliable operation, if the system has otherwise gobs of RAM, which just don't happen to be atomic allocatable!
Quoted 1 line
[...] With
GFP_KERNELwe might hang the system.
Yes, that's a bug that should be fixed - but not via GFP_ATOMIC, which adds insult to injury.
Thanks,
Ingo
Quoted 44 lines · Ingo Molnar
Quoted 41 lines · Prarit Bhargava
Quoted 39 lines · Ingo Molnar
Quoted 24 lines · Yinghai Lu
Fix warning:
arch/x86/kernel/irq.c: In functioncheck_irq_vectors_for_cpu_disable:arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052 bytes is larger than 2048 byteswhen NR_CPUS=8192
We should use
zalloc_cpumask_var()instead.-v2: update to
GFP_ATOMICinstead and free the allocated cpumask at last.Signed-off-by Yinghai Lu <[email protected]>Cc Prarit Bhargava <[email protected]>--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)Index:
linux-2.6/arch/x86/kernel/irq.c
===================================================================arch/x86/kernel/irq.c +0
-0@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;arch/x86/kernel/irq.c int check_irq_vectors_for_cpu_disable(vo · line 287
Quoted 10 lines
arch/x86/kernel/irq.c +8
-1struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; + +if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + }Atomic allocations can fail easily if the system is under duress.
Quoted 5 lines
Quoted 1 line
Then the hotplug attempt will fail which IMO is okay. [...]
Which is not OK at all for reliable operation, if the system has otherwise gobs of RAM, which just don't happen to be atomic allocatable!
Ingo, I'm really not sure what other option there is here. Care to suggest one?
P.
Quoted 47 lines · Prarit Bhargava
Quoted 44 lines · Ingo Molnar
Quoted 41 lines · Prarit Bhargava
Quoted 39 lines · Ingo Molnar
Quoted 24 lines · Yinghai Lu
Fix warning:
arch/x86/kernel/irq.c: In functioncheck_irq_vectors_for_cpu_disable:arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052 bytes is larger than 2048 byteswhen NR_CPUS=8192
We should use
zalloc_cpumask_var()instead.-v2: update to
GFP_ATOMICinstead and free the allocated cpumask at last.Signed-off-by Yinghai Lu <[email protected]>Cc Prarit Bhargava <[email protected]>--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)Index:
linux-2.6/arch/x86/kernel/irq.c
===================================================================arch/x86/kernel/irq.c +0
-0@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;arch/x86/kernel/irq.c int check_irq_vectors_for_cpu_disable(vo · line 287
Quoted 10 lines
arch/x86/kernel/irq.c +8
-1struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; + +if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + }Atomic allocations can fail easily if the system is under duress.
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
Then the hotplug attempt will fail which IMO is okay. [...]
Which is not OK at all for reliable operation, if the system has otherwise gobs of RAM, which just don't happen to be atomic allocatable!
Ingo, I'm really not sure what other option there is here. Care to suggest one?
Since only ever a single instance of this code will run, can it simply be a global cpumask_t variable?
Thanks,
Ingo
H. Peter Anvin
reviewers/global/static/, with a big loud comment why it is okay.
Quoted 69 lines · Ingo Molnar
Quoted 4 lines · Prarit Bhargava
Quoted 1 line · Ingo Molnar
Quoted 8 lines
Quoted 8 lines · Prarit Bhargava
Quoted 7 lines · Ingo Molnar
Quoted 5 lines · Yinghai Lu
Quoted 3 lines
Fix warning:
arch/x86/kernel/irq.c: In function
check_irq_vectors_for_cpu_disable:Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
arch/x86/kernel/irq.c:337:1: warning: the frame size of 2052bytes is larger than 2048 bytes
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
when NR_CPUS=8192
We should use
zalloc_cpumask_var()instead.-v2: update to
GFP_ATOMICinstead and free the allocated cpumaskat last.
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Signed-off-by Yinghai Lu <[email protected]>Cc Prarit Bhargava <[email protected]>--- arch/x86/kernel/irq.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-)Index:
linux-2.6/arch/x86/kernel/irq.c===================================================================
Quoted 18 lines
Quoted 18 lines
Quoted 18 lines
Quoted 17 lines
Quoted 5 lines
arch/x86/kernel/irq.c +0
-0@@ -277,11 +277,18 @@ int check_irq_vectors_for_cpu_disable(vounsigned int this_cpu, vector, this_count, count;struct irq_desc *desc;arch/x86/kernel/irq.c int check_irq_vectors_for_cpu_disable(vo · line 287
Quoted 10 lines
arch/x86/kernel/irq.c +8
-1struct irq_data *data; -struct cpumask affinity_new, online_new; +cpumask_var_t affinity_new, online_new; + +if (!alloc_cpumask_var (&affinity_new,GFP_ATOMIC )) +return -ENOMEM ; +if (!alloc_cpumask_var (&online_new,GFP_ATOMIC )) { +free_cpumask_var (affinity_new); +return -ENOMEM ; + }Atomic allocations can fail easily if the system is under duress.
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
Then the hotplug attempt will fail which IMO is okay. [...]
Which is not OK at all for reliable operation, if the system has otherwise gobs of RAM, which just don't happen to be atomic allocatable!
Ingo, I'm really not sure what other option there is here. Care to suggest one?
Since only ever a single instance of this code will run, can it simply be a global
cpumask_tvariable?Thanks,
Ingo
Signature · 1 line
Sent from my mobile phone. Please pardon brevity and lack of formatting.
Quoted 1 line · H. Peter Anvin
s/global/static/, with a big loud comment why it is okay.
It would be a global no matter which form we use, but for maintainability reasons I generally prefer a static put right before the function that uses it:
static cpumask_t mask;
static func(...)
{
}That makes it really apparent that it's a global - statics are easily missed when hiding amongst local variables.
Thanks,
Ingo
H. Peter Anvin
reviewerI strongly disagree with putting variables in file scope when function scope will do, but I do like to see static variables before automatics. Anyway, this is bikeshedding.
Quoted 21 lines · Ingo Molnar
Quoted 1 line · H. Peter Anvin
s/global/static/, with a big loud comment why it is okay.
It would be a global no matter which form we use, but for maintainability reasons I generally prefer a static put right before the function that uses it:
static cpumask_t mask; static func(...) { }That makes it really apparent that it's a global - statics are easily missed when hiding amongst local variables.
Thanks,
Ingo
Signature · 1 line
Sent from my mobile phone. Please pardon brevity and lack of formatting.
Quoted 20 lines · H. Peter Anvin
I strongly disagree with putting variables in file scope when function scope will do, but I do like to see static variables before automatics. Anyway, this is bikeshedding.
Quoted 17 lines · Ingo Molnar
Quoted 1 line · H. Peter Anvin
s/global/static/, with a big loud comment why it is okay.
It would be a global no matter which form we use, but for maintainability reasons I generally prefer a static put right before the function that uses it:
static cpumask_t mask; static func(...) { }That makes it really apparent that it's a global - statics are easily missed when hiding amongst local variables.
Okay, thanks for the input -- I'll put something together and test.
P.
Quoted 5 lines
Quoted 4 lines
Thanks,
Ingo
Quoted 2 lines · H. Peter Anvin
I strongly disagree with putting variables in file scope when function scope will do, [...]
Yes, you are right that single-use file scope statics 'could' be moved function local and are syntactically superior because in that case other functions cannot make use of it.
But I also have very good (and unfixable and thus stronger) reasons to object to statics inside local variables: more than once I personally missed 'hidden statics' during review, in one case it even slipped into a commit, so it's not a practice I want to encourage in any shape or form (even if the 'rule' is to have a big fat comment, people will just see the function local static and emulate it without the comment), for code I maintain.
It's not about you, it's about me and other reviewers: I've seen statics slipping past other reviewers as well. So it's the lesser of two evils. Can you accept that reasoning?
Thanks,
Ingo