* [PATCH 0/3] powerpc/64s/radix: misc tidying
@ 2023-02-03 11:17 Nicholas Piggin
2023-02-03 11:17 ` [PATCH 1/3] powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush Nicholas Piggin
` (3 more replies)
0 siblings, 4 replies; 5+ messages in thread
From: Nicholas Piggin @ 2023-02-03 11:17 UTC (permalink / raw)
To: linuxppc-dev; +Cc: Nicholas Piggin
Here's a few things I wanted to tidy before doing the lazy tlb mm
shootdown / exit TLB flush IPI optimisation. There is no hard
dependency just slight code rearrangement but since we're waiting
for the shootdown patches to go via the -mm tree we could do these
now.
It basically just removing a bunch of dead code, or replacing them
with WARN so we can remove them entirely in a release or two.
Thanks,
Nick
Nicholas Piggin (3):
powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush
powerpc/64s/radix: mm->context.id should always be valid
powerpc/64s/radix: Remove TLB_FLUSH_ALL test from range flushes
arch/powerpc/mm/book3s64/radix_tlb.c | 62 ++++++++++++++--------------
1 file changed, 32 insertions(+), 30 deletions(-)
--
2.37.2
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH 1/3] powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush
2023-02-03 11:17 [PATCH 0/3] powerpc/64s/radix: misc tidying Nicholas Piggin
@ 2023-02-03 11:17 ` Nicholas Piggin
2023-02-03 11:17 ` [PATCH 2/3] powerpc/64s/radix: mm->context.id should always be valid Nicholas Piggin
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: Nicholas Piggin @ 2023-02-03 11:17 UTC (permalink / raw)
To: linuxppc-dev; +Cc: Nicholas Piggin
need_flush_all is only set by arch code to instruct generic tlb_flush
to flush all. It is never set by powerpc, so it can be removed.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
arch/powerpc/mm/book3s64/radix_tlb.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/powerpc/mm/book3s64/radix_tlb.c b/arch/powerpc/mm/book3s64/radix_tlb.c
index 282359ab525b..1e451184cba4 100644
--- a/arch/powerpc/mm/book3s64/radix_tlb.c
+++ b/arch/powerpc/mm/book3s64/radix_tlb.c
@@ -1302,7 +1302,7 @@ void radix__tlb_flush(struct mmu_gather *tlb)
* that flushes the process table entry cache upon process teardown.
* See the comment for radix in arch_exit_mmap().
*/
- if (tlb->fullmm || tlb->need_flush_all) {
+ if (tlb->fullmm) {
__flush_all_mm(mm, true);
} else if ( (psize = radix_get_mmu_psize(page_size)) == -1) {
if (!tlb->freed_tables)
--
2.37.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH 2/3] powerpc/64s/radix: mm->context.id should always be valid
2023-02-03 11:17 [PATCH 0/3] powerpc/64s/radix: misc tidying Nicholas Piggin
2023-02-03 11:17 ` [PATCH 1/3] powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush Nicholas Piggin
@ 2023-02-03 11:17 ` Nicholas Piggin
2023-02-03 11:17 ` [PATCH 3/3] powerpc/64s/radix: Remove TLB_FLUSH_ALL test from range flushes Nicholas Piggin
2023-02-15 12:40 ` [PATCH 0/3] powerpc/64s/radix: misc tidying Michael Ellerman
3 siblings, 0 replies; 5+ messages in thread
From: Nicholas Piggin @ 2023-02-03 11:17 UTC (permalink / raw)
To: linuxppc-dev; +Cc: Nicholas Piggin
The MMU_NO_CONTEXT checks are an unnecessary complication. Make
these warn to prepare to remove them in future.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
arch/powerpc/mm/book3s64/radix_tlb.c | 39 +++++++++++++++-------------
1 file changed, 21 insertions(+), 18 deletions(-)
diff --git a/arch/powerpc/mm/book3s64/radix_tlb.c b/arch/powerpc/mm/book3s64/radix_tlb.c
index 1e451184cba4..27aa5b0fbb3f 100644
--- a/arch/powerpc/mm/book3s64/radix_tlb.c
+++ b/arch/powerpc/mm/book3s64/radix_tlb.c
@@ -700,12 +700,13 @@ static inline void _tlbiel_va_range_multicast(struct mm_struct *mm,
*/
void radix__local_flush_tlb_mm(struct mm_struct *mm)
{
- unsigned long pid;
+ unsigned long pid = mm->context.id;
+
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
+ return;
preempt_disable();
- pid = mm->context.id;
- if (pid != MMU_NO_CONTEXT)
- _tlbiel_pid(pid, RIC_FLUSH_TLB);
+ _tlbiel_pid(pid, RIC_FLUSH_TLB);
preempt_enable();
}
EXPORT_SYMBOL(radix__local_flush_tlb_mm);
@@ -713,12 +714,13 @@ EXPORT_SYMBOL(radix__local_flush_tlb_mm);
#ifndef CONFIG_SMP
void radix__local_flush_all_mm(struct mm_struct *mm)
{
- unsigned long pid;
+ unsigned long pid = mm->context.id;
+
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
+ return;
preempt_disable();
- pid = mm->context.id;
- if (pid != MMU_NO_CONTEXT)
- _tlbiel_pid(pid, RIC_FLUSH_ALL);
+ _tlbiel_pid(pid, RIC_FLUSH_ALL);
preempt_enable();
}
EXPORT_SYMBOL(radix__local_flush_all_mm);
@@ -732,12 +734,13 @@ static void __flush_all_mm(struct mm_struct *mm, bool fullmm)
void radix__local_flush_tlb_page_psize(struct mm_struct *mm, unsigned long vmaddr,
int psize)
{
- unsigned long pid;
+ unsigned long pid = mm->context.id;
+
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
+ return;
preempt_disable();
- pid = mm->context.id;
- if (pid != MMU_NO_CONTEXT)
- _tlbiel_va(vmaddr, pid, psize, RIC_FLUSH_TLB);
+ _tlbiel_va(vmaddr, pid, psize, RIC_FLUSH_TLB);
preempt_enable();
}
@@ -945,7 +948,7 @@ void radix__flush_tlb_mm(struct mm_struct *mm)
enum tlb_flush_type type;
pid = mm->context.id;
- if (unlikely(pid == MMU_NO_CONTEXT))
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
preempt_disable();
@@ -985,7 +988,7 @@ static void __flush_all_mm(struct mm_struct *mm, bool fullmm)
enum tlb_flush_type type;
pid = mm->context.id;
- if (unlikely(pid == MMU_NO_CONTEXT))
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
preempt_disable();
@@ -1024,7 +1027,7 @@ void radix__flush_tlb_page_psize(struct mm_struct *mm, unsigned long vmaddr,
enum tlb_flush_type type;
pid = mm->context.id;
- if (unlikely(pid == MMU_NO_CONTEXT))
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
preempt_disable();
@@ -1130,7 +1133,7 @@ static inline void __radix__flush_tlb_range(struct mm_struct *mm,
enum tlb_flush_type type;
pid = mm->context.id;
- if (unlikely(pid == MMU_NO_CONTEXT))
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
preempt_disable();
@@ -1330,7 +1333,7 @@ static void __radix__flush_tlb_range_psize(struct mm_struct *mm,
enum tlb_flush_type type;
pid = mm->context.id;
- if (unlikely(pid == MMU_NO_CONTEXT))
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
fullmm = (end == TLB_FLUSH_ALL);
@@ -1406,7 +1409,7 @@ void radix__flush_tlb_collapsed_pmd(struct mm_struct *mm, unsigned long addr)
enum tlb_flush_type type;
pid = mm->context.id;
- if (unlikely(pid == MMU_NO_CONTEXT))
+ if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
/* 4k page size, just blow the world */
--
2.37.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* [PATCH 3/3] powerpc/64s/radix: Remove TLB_FLUSH_ALL test from range flushes
2023-02-03 11:17 [PATCH 0/3] powerpc/64s/radix: misc tidying Nicholas Piggin
2023-02-03 11:17 ` [PATCH 1/3] powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush Nicholas Piggin
2023-02-03 11:17 ` [PATCH 2/3] powerpc/64s/radix: mm->context.id should always be valid Nicholas Piggin
@ 2023-02-03 11:17 ` Nicholas Piggin
2023-02-15 12:40 ` [PATCH 0/3] powerpc/64s/radix: misc tidying Michael Ellerman
3 siblings, 0 replies; 5+ messages in thread
From: Nicholas Piggin @ 2023-02-03 11:17 UTC (permalink / raw)
To: linuxppc-dev; +Cc: Nicholas Piggin
This looks like it came across from x86, but x86 uses TLB_FLUSH_ALL as
a parameter to internal functions. Powerpc never sets it anywhere.
Remove the associated logic and leave a warning for now.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
arch/powerpc/mm/book3s64/radix_tlb.c | 21 ++++++++++-----------
1 file changed, 10 insertions(+), 11 deletions(-)
diff --git a/arch/powerpc/mm/book3s64/radix_tlb.c b/arch/powerpc/mm/book3s64/radix_tlb.c
index 27aa5b0fbb3f..2608ff7cf62c 100644
--- a/arch/powerpc/mm/book3s64/radix_tlb.c
+++ b/arch/powerpc/mm/book3s64/radix_tlb.c
@@ -1107,6 +1107,9 @@ void radix__flush_tlb_kernel_range(unsigned long start, unsigned long end)
}
EXPORT_SYMBOL(radix__flush_tlb_kernel_range);
+/*
+ * Doesn't appear to be used anywhere. Remove.
+ */
#define TLB_FLUSH_ALL -1UL
/*
@@ -1128,7 +1131,6 @@ static inline void __radix__flush_tlb_range(struct mm_struct *mm,
unsigned int page_shift = mmu_psize_defs[mmu_virtual_psize].shift;
unsigned long page_size = 1UL << page_shift;
unsigned long nr_pages = (end - start) >> page_shift;
- bool fullmm = (end == TLB_FLUSH_ALL);
bool flush_pid, flush_pwc = false;
enum tlb_flush_type type;
@@ -1136,15 +1138,15 @@ static inline void __radix__flush_tlb_range(struct mm_struct *mm,
if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
+ WARN_ON_ONCE(end == TLB_FLUSH_ALL);
+
preempt_disable();
smp_mb(); /* see radix__flush_tlb_mm */
- type = flush_type_needed(mm, fullmm);
+ type = flush_type_needed(mm, false);
if (type == FLUSH_TYPE_NONE)
goto out;
- if (fullmm)
- flush_pid = true;
- else if (type == FLUSH_TYPE_GLOBAL)
+ if (type == FLUSH_TYPE_GLOBAL)
flush_pid = nr_pages > tlb_single_page_flush_ceiling;
else
flush_pid = nr_pages > tlb_local_single_page_flush_ceiling;
@@ -1328,7 +1330,6 @@ static void __radix__flush_tlb_range_psize(struct mm_struct *mm,
unsigned int page_shift = mmu_psize_defs[psize].shift;
unsigned long page_size = 1UL << page_shift;
unsigned long nr_pages = (end - start) >> page_shift;
- bool fullmm = (end == TLB_FLUSH_ALL);
bool flush_pid;
enum tlb_flush_type type;
@@ -1336,17 +1337,15 @@ static void __radix__flush_tlb_range_psize(struct mm_struct *mm,
if (WARN_ON_ONCE(pid == MMU_NO_CONTEXT))
return;
- fullmm = (end == TLB_FLUSH_ALL);
+ WARN_ON_ONCE(end == TLB_FLUSH_ALL);
preempt_disable();
smp_mb(); /* see radix__flush_tlb_mm */
- type = flush_type_needed(mm, fullmm);
+ type = flush_type_needed(mm, false);
if (type == FLUSH_TYPE_NONE)
goto out;
- if (fullmm)
- flush_pid = true;
- else if (type == FLUSH_TYPE_GLOBAL)
+ if (type == FLUSH_TYPE_GLOBAL)
flush_pid = nr_pages > tlb_single_page_flush_ceiling;
else
flush_pid = nr_pages > tlb_local_single_page_flush_ceiling;
--
2.37.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH 0/3] powerpc/64s/radix: misc tidying
2023-02-03 11:17 [PATCH 0/3] powerpc/64s/radix: misc tidying Nicholas Piggin
` (2 preceding siblings ...)
2023-02-03 11:17 ` [PATCH 3/3] powerpc/64s/radix: Remove TLB_FLUSH_ALL test from range flushes Nicholas Piggin
@ 2023-02-15 12:40 ` Michael Ellerman
3 siblings, 0 replies; 5+ messages in thread
From: Michael Ellerman @ 2023-02-15 12:40 UTC (permalink / raw)
To: linuxppc-dev, Nicholas Piggin
On Fri, 3 Feb 2023 21:17:15 +1000, Nicholas Piggin wrote:
> Here's a few things I wanted to tidy before doing the lazy tlb mm
> shootdown / exit TLB flush IPI optimisation. There is no hard
> dependency just slight code rearrangement but since we're waiting
> for the shootdown patches to go via the -mm tree we could do these
> now.
>
> It basically just removing a bunch of dead code, or replacing them
> with WARN so we can remove them entirely in a release or two.
>
> [...]
Applied to powerpc/next.
[1/3] powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush
https://git.kernel.org/powerpc/c/45abf5d94b9bd0eebca5c7272788e2d16c8b5b43
[2/3] powerpc/64s/radix: mm->context.id should always be valid
https://git.kernel.org/powerpc/c/d01dc25e47af9d30185ca12bb9e221d6af915d9f
[3/3] powerpc/64s/radix: Remove TLB_FLUSH_ALL test from range flushes
https://git.kernel.org/powerpc/c/dcfecb989afdf9101ee42e2adf04756a2ea4819d
cheers
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2023-02-15 12:49 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-02-03 11:17 [PATCH 0/3] powerpc/64s/radix: misc tidying Nicholas Piggin
2023-02-03 11:17 ` [PATCH 1/3] powerpc/64s/radix: Remove need_flush_all test from radix__tlb_flush Nicholas Piggin
2023-02-03 11:17 ` [PATCH 2/3] powerpc/64s/radix: mm->context.id should always be valid Nicholas Piggin
2023-02-03 11:17 ` [PATCH 3/3] powerpc/64s/radix: Remove TLB_FLUSH_ALL test from range flushes Nicholas Piggin
2023-02-15 12:40 ` [PATCH 0/3] powerpc/64s/radix: misc tidying Michael Ellerman
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).