"Aneesh Kumar K.V" <aneesh.ku...@linux.ibm.com> writes:
> Similar to x86/s390 add a debugfs file to tune tlb_single_page_flush_ceiling.
> Also add a debugfs entry for tlb_local_single_page_flush_ceiling.
>
> Signed-off-by: Aneesh Kumar K.V <aneesh.ku...@linux.ibm.com>
> ---
> Changes from v1:
> * switch to debugfs_create_u32
>
> arch/powerpc/mm/book3s64/radix_tlb.c | 16 ++++++++++++++--
> 1 file changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/arch/powerpc/mm/book3s64/radix_tlb.c
> b/arch/powerpc/mm/book3s64/radix_tlb.c
> index aefc100d79a7..1fa2bc6a969e 100644
> --- a/arch/powerpc/mm/book3s64/radix_tlb.c
> +++ b/arch/powerpc/mm/book3s64/radix_tlb.c
> @@ -17,6 +17,7 @@
> #include <asm/trace.h>
> #include <asm/cputhreads.h>
> #include <asm/plpar_wrappers.h>
> +#include <asm/debugfs.h>
>
> #include "internal.h"
>
> @@ -1106,8 +1107,8 @@ EXPORT_SYMBOL(radix__flush_tlb_kernel_range);
> * invalidating a full PID, so it has a far lower threshold to change from
> * individual page flushes to full-pid flushes.
> */
> -static unsigned long tlb_single_page_flush_ceiling __read_mostly = 33;
> -static unsigned long tlb_local_single_page_flush_ceiling __read_mostly =
> POWER9_TLB_SETS_RADIX * 2;
> +static u32 tlb_single_page_flush_ceiling __read_mostly = 33;
> +static u32 tlb_local_single_page_flush_ceiling __read_mostly =
> POWER9_TLB_SETS_RADIX * 2;
>
> static inline void __radix__flush_tlb_range(struct mm_struct *mm,
> unsigned long start, unsigned long
> end)
> @@ -1524,3 +1525,14 @@ void do_h_rpt_invalidate_prt(unsigned long pid,
> unsigned long lpid,
> EXPORT_SYMBOL_GPL(do_h_rpt_invalidate_prt);
>
> #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
> +
> +static int __init create_tlb_single_page_flush_ceiling(void)
> +{
> + debugfs_create_u32("tlb_single_page_flush_ceiling", 0600,
> + powerpc_debugfs_root,
> &tlb_single_page_flush_ceiling);
> + debugfs_create_u32("tlb_local_single_page_flush_ceiling", 0600,
> + powerpc_debugfs_root,
> &tlb_local_single_page_flush_ceiling);
> + return 0;
> +}
> +late_initcall(create_tlb_single_page_flush_ceiling);
This patch seems to do what the commit message says, and it does seem to
make sense to have these parameters as tunables.
I was briefly concerned that switching from an unsigned long to a u32
might lead to suboptimal code generation in older gcc versions, but it
doesn't seem to be a case where a single instruction is going to make a
huge impact.
I also wondered what the C integer promotion rules would do with a the
nr_pages > tlb*flush_ceiling comparisons, but if we are trying to flush
more than 4 billion pages we might have other, bigger problems! (Also,
if I understand the C integer rules correctly the u32 will get promoted
to an unsigned long anyway.)
All in all this seems good to me.
Reviewed-by: Daniel Axtens <d...@axtens.net>
Kind regards,
Daniel
> +
> --
> 2.31.1
