On Wed, Jan 15, 2020 at 08:52:45PM +0000, Chris Wilson wrote:
> Since we may try and flush the cachelines associated with large buffers
> (an 8K framebuffer is about 128MiB, even before we try HDR), this leads
> to unacceptably long latencies (when using a voluntary CONFIG_PREEMPT).
> If we call cond_resched() between each sg chunk, that it about every 128
> pages, we have a natural break point in which to check if the process
> needs to be rescheduled. Naturally, this means that drm_clflush_sg() can
> only be called from process context -- which is true at the moment. The
> other clflush routines remain usable from atomic context.
>
> Even though flushing large objects takes a demonstrable amount to time
> to flush all the cachelines, clflush is still preferred over a
> system-wide wbinvd as the latter has unpredictable latencies affecting
> the whole system not just the local task.
>
> Reported-by: David Laight <David.Laight@xxxxxxxxxx>
> Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> Cc: David Laight <David.Laight@xxxxxxxxxx>
The original bug report is complaining about latencies for SCHED_RT
threads, on a system that doesn't even use CONFIG_PREEMPT. I'm not sure
it's terribly valid to cater to that use-case - all the desktop distros
seem a lot more reasonable. So firmly *shrug* from my side ...
Patch itself looks correct, just not seeing the point.
-Daniel
> ---
> drivers/gpu/drm/drm_cache.c | 49 ++++++++++++++++++++++++++++++++++---
> 1 file changed, 45 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/drm_cache.c b/drivers/gpu/drm/drm_cache.c
> index 03e01b000f7a..fbd2bb644544 100644
> --- a/drivers/gpu/drm/drm_cache.c
> +++ b/drivers/gpu/drm/drm_cache.c
> @@ -112,23 +112,64 @@ drm_clflush_pages(struct page *pages[], unsigned long num_pages)
> }
> EXPORT_SYMBOL(drm_clflush_pages);
>
> +static __always_inline struct sgt_iter {
> + struct scatterlist *sgp;
> + unsigned long pfn;
> + unsigned int curr;
> + unsigned int max;
> +} __sgt_iter(struct scatterlist *sgl) {
> + struct sgt_iter s = { .sgp = sgl };
> +
> + if (s.sgp) {
> + s.max = s.curr = s.sgp->offset;
> + s.max += s.sgp->length;
> + s.pfn = page_to_pfn(sg_page(s.sgp));
> + }
> +
> + return s;
> +}
> +
> +static inline struct scatterlist *__sg_next_resched(struct scatterlist *sg)
> +{
> + if (sg_is_last(sg))
> + return NULL;
> +
> + ++sg;
> + if (unlikely(sg_is_chain(sg))) {
> + sg = sg_chain_ptr(sg);
> + cond_resched();
> + }
> + return sg;
> +}
> +
> +#define for_each_sgt_page(__pp, __iter, __sgt) \
> + for ((__iter) = __sgt_iter((__sgt)->sgl); \
> + ((__pp) = (__iter).pfn == 0 ? NULL : \
> + pfn_to_page((__iter).pfn + ((__iter).curr >> PAGE_SHIFT))); \
> + (((__iter).curr += PAGE_SIZE) >= (__iter).max) ? \
> + (__iter) = __sgt_iter(__sg_next_resched((__iter).sgp)), 0 : 0)
> +
> /**
> * drm_clflush_sg - Flush dcache lines pointing to a scather-gather.
> * @st: struct sg_table.
> *
> * Flush every data cache line entry that points to an address in the
> - * sg.
> + * sg. This may schedule between scatterlist chunks, in order to keep
> + * the system preemption-latency down for large buffers.
> */
> void
> drm_clflush_sg(struct sg_table *st)
> {
> + might_sleep();
> +
> #if defined(CONFIG_X86)
> if (static_cpu_has(X86_FEATURE_CLFLUSH)) {
> - struct sg_page_iter sg_iter;
> + struct sgt_iter sg_iter;
> + struct page *page;
>
> mb(); /*CLFLUSH is ordered only by using memory barriers*/
> - for_each_sg_page(st->sgl, &sg_iter, st->nents, 0)
> - drm_clflush_page(sg_page_iter_page(&sg_iter));
> + for_each_sgt_page(page, sg_iter, st)
> + drm_clflush_page(page);
> mb(); /*Make sure that all cache line entry is flushed*/
>
> return;
> --
> 2.25.0
>
> _______________________________________________
> dri-devel mailing list
> dri-devel@xxxxxxxxxxxxxxxxxxxxx
> https://lists.freedesktop.org/mailman/listinfo/dri-devel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
