On Mon, Jul 27, 2020 at 09:41:14PM -0700, Ricardo Neri wrote:
> I think I got a little lost here.
Hehe, sorry. I got carried away, it's just that recently people
expressed interest in 'fixing' some of the text_poke_sync() issues
again.
> If I understand correctly, there are
> two alternatives to
On Mon, Jul 27, 2020 at 03:30:20PM +0200, pet...@infradead.org wrote:
> On Mon, Jul 27, 2020 at 03:05:36PM +0200, pet...@infradead.org wrote:
> > Yeah, I'm not sure.. the 'funny' thing is that typically call
> > sync_core() from an IPI anyway. And the synchronous broadcast IPI is by
> > far the mos
On Mon, Jul 27, 2020 at 05:47:32AM -0700, h...@zytor.com wrote:
> On July 27, 2020 1:20:03 AM PDT, pet...@infradead.org wrote:
> >On Sun, Jul 26, 2020 at 09:31:32PM -0700, Ricardo Neri wrote:
> >> +static inline void serialize(void)
> >> +{
> >> + asm volatile(".byte 0xf, 0x1, 0xe8");
> >> +}
> >
On Mon, Jul 27, 2020 at 10:20:03AM +0200, pet...@infradead.org wrote:
> On Sun, Jul 26, 2020 at 09:31:32PM -0700, Ricardo Neri wrote:
> > +static inline void serialize(void)
> > +{
> > + asm volatile(".byte 0xf, 0x1, 0xe8");
> > +}
>
> Can we pretty please have a comment with the binutils versio
> For a really overenginered solution, but which might perform unnecessary
> poorly on existing hardware:
>
> asm volatile("1: .byte 0xf, 0x1, 0xe8; 2:"
>_ASM_EXTABLE(1b,2b));
You win the prize for the smallest code. Might need (the already large)
comment to double
in si
On Mon, Jul 27, 2020 at 03:05:36PM +0200, pet...@infradead.org wrote:
> Yeah, I'm not sure.. the 'funny' thing is that typically call
> sync_core() from an IPI anyway. And the synchronous broadcast IPI is by
> far the most expensive part of that.
>
> Something like this...
>
> diff --git a/arch/x
On Mon, Jul 27, 2020 at 05:49:28AM -0700, h...@zytor.com wrote:
> On July 27, 2020 1:36:19 AM PDT, pet...@infradead.org wrote:
> >On Sun, Jul 26, 2020 at 10:55:15PM -0700, h...@zytor.com wrote:
> >> For a really overenginered solution, but which might perform
> >> unnecessary poorly on existing har
On July 27, 2020 1:36:19 AM PDT, pet...@infradead.org wrote:
>On Sun, Jul 26, 2020 at 10:55:15PM -0700, h...@zytor.com wrote:
>> For a really overenginered solution, but which might perform
>> unnecessary poorly on existing hardware:
>>
>> asm volatile("1: .byte 0xf, 0x1, 0xe8; 2:"
>>
On July 27, 2020 1:20:03 AM PDT, pet...@infradead.org wrote:
>On Sun, Jul 26, 2020 at 09:31:32PM -0700, Ricardo Neri wrote:
>> +static inline void serialize(void)
>> +{
>> +asm volatile(".byte 0xf, 0x1, 0xe8");
>> +}
>
>Can we pretty please have a comment with the binutils version that has
>the
On Sun, Jul 26, 2020 at 10:55:15PM -0700, h...@zytor.com wrote:
> For a really overenginered solution, but which might perform
> unnecessary poorly on existing hardware:
>
> asm volatile("1: .byte 0xf, 0x1, 0xe8; 2:"
> _ASM_EXTABLE(1b,2b));
Ha! cute, you take an #UD ?
We
On Sun, Jul 26, 2020 at 09:31:32PM -0700, Ricardo Neri wrote:
> @@ -75,6 +77,12 @@ static inline void sync_core(void)
>* Like all of Linux's memory ordering operations, this is a
>* compiler barrier as well.
>*/
> +
> + if (boot_cpu_has(X86_FEATURE_SERIALIZE)) {
> +
On Sun, Jul 26, 2020 at 09:31:32PM -0700, Ricardo Neri wrote:
> +static inline void serialize(void)
> +{
> + asm volatile(".byte 0xf, 0x1, 0xe8");
> +}
Can we pretty please have a comment with the binutils version that has
the mnomic? Such that when we increase the required binutils version we
On July 26, 2020 10:55:15 PM PDT, h...@zytor.com wrote:
>On July 26, 2020 9:31:32 PM PDT, Ricardo Neri
> wrote:
>>The SERIALIZE instruction gives software a way to force the processor
>>to
>>complete all modifications to flags, registers and memory from
>previous
>>instructions and drain all buffer
On July 26, 2020 9:31:32 PM PDT, Ricardo Neri
wrote:
>The SERIALIZE instruction gives software a way to force the processor
>to
>complete all modifications to flags, registers and memory from previous
>instructions and drain all buffered writes to memory before the next
>instruction is fetched an
The SERIALIZE instruction gives software a way to force the processor to
complete all modifications to flags, registers and memory from previous
instructions and drain all buffered writes to memory before the next
instruction is fetched and executed. Thus, it serves the purpose of
sync_core(). Use
15 matches
Mail list logo
