Menu
▾
▴
[linuxsh-dev] [PATCH] SH4 Cache Routine Optimizations
|
From: Paul M. <le...@Ch...> - 2002-02-28 07:47:24
|
Hello,
These are a small set of cleanups for the SH4 caches that have been sitting=
in
my home directory for a little while, and after having just recently reciev=
ed
testing, have been deemed permissible to see the light of day.
The changes are pretty simplistic, a few sanity checks here and there to av=
oid
some wasted cache flushes.. as well as a proper flush_cache_range()
implementation that doesn't just act as a dummy wrapper to flush_cache_all(=
).
Seems to be holding up so far.
With what testing this patch has recieved thus far (thanks to M. R. Brown),
it's been a bit quicker in the majority of cases..=20
Writeback -
original Lethal's=20
tunnel 13.4169fps 13.4272fps
flamme 12.6753fps 12.6564fps -
dawafire 9.3169fps 9.300047fps -
bump 10.8493fps 10.8635fps
rotozoom 12.0676fps 12.054fps -
plasma 8.33784fps 8.37143fps
sdl/testwin 2175ms 2111ms
sdl/testsprite 29.88fps 30.04fps
sdl/testpalette 531/15.06fps 673/15.09fps
These are obviously not very concrete benchmarks, and perhaps a bit of kern=
el
profiling is called for if anyone is really hesitant about any of the chang=
es.
Though I'd like to avoid writing a new test suite for stressing caches if at
all possible..
These changes are of course only related to SH4 caches configured in
write-back mode.. I've started on a patch for write-through, but that's sti=
ll
a work-in-progress.
Regards,
--=20
Paul Mundt <le...@ch...>
Index: ChangeLog
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
RCS file: /cvsroot/linuxsh/linux/ChangeLog,v
retrieving revision 1.32
diff -u -r1.32 ChangeLog
--- ChangeLog 26 Feb 2002 09:15:15 -0000 1.32
+++ ChangeLog 28 Feb 2002 07:34:14 -0000
@@ -1,3 +1,9 @@
+2002-02-27 Paul Mundt <le...@ch...>
+
+ * arch/sh/mm/cache-sh4.c (__flush_icache_page): Added back in.
+ (flush_cache_range): Rewritten entirely.
+ (flush_cache_mm): Only flush caches on a valid mm context.
+
2002-02-26 NIIBE Yutaka <gn...@m1...>
=20
Updated to 2.5.3.
Index: arch/sh/mm/cache-sh4.c
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
RCS file: /cvsroot/linuxsh/linux/arch/sh/mm/cache-sh4.c,v
retrieving revision 1.1.1.1
diff -u -r1.1.1.1 cache-sh4.c
--- arch/sh/mm/cache-sh4.c 15 Oct 2001 20:44:53 -0000 1.1.1.1
+++ arch/sh/mm/cache-sh4.c 25 Feb 2002 08:43:04 -0000
@@ -3,7 +3,7 @@
* linux/arch/sh/mm/cache.c
*
* Copyright (C) 1999, 2000 Niibe Yutaka
- *
+ * Copyright (C) 2001, 2002 Paul Mundt
*/
=20
#include <linux/config.h>
@@ -109,7 +114,7 @@
unsigned long size, unsigned long flags);
=20
if (remap_area_pages(P3SEG, 0, PAGE_SIZE*4, _PAGE_CACHABLE))
- panic("p3_cachie_init failed.");
+ panic("%s failed.", __FUNCTION__);
sema_init (&p3map_sem[0], 1);
sema_init (&p3map_sem[1], 1);
sema_init (&p3map_sem[2], 1);
@@ -231,19 +236,30 @@
ctrl_outl(0, addr);
}
=20
-#if 0 /* DEBUG DEBUG */
+ back_to_P1();
+ restore_flags(flags);
+}
+
+static void __flush_icache_page(unsigned long phys)
+{
+ unsigned long addr, data;
+ unsigned long flags;
+
+ phys |=3D CACHE_VALID;
+
+ save_and_cli(flags);
+ jump_to_P2();
+
/* Loop all the I-cache */
for (addr =3D CACHE_IC_ADDRESS_ARRAY;
addr < (CACHE_IC_ADDRESS_ARRAY
+(CACHE_IC_NUM_ENTRIES<< CACHE_IC_ENTRY_SHIFT));
addr +=3D (1<<CACHE_IC_ENTRY_SHIFT)) {
data =3D ctrl_inl(addr)&(0x1ffff000|CACHE_VALID);
- if (data =3D=3D phys) {
- printk(KERN_INFO "__flush_cache_page: I-cache entry found\n");
+ if (data =3D=3D phys)
ctrl_outl(0, addr);
- }
}
-#endif
+
back_to_P1();
restore_flags(flags);
}
@@ -283,7 +299,19 @@
{
/* Is there any good way? */
/* XXX: possibly call flush_cache_range for each vm area */
- flush_cache_all();
+ /*=20
+ * FIXME: Really, the optimal solution here would be able to flush out
+ * individual lines created by the specified context, but this isn't
+ * feasible for a number of architectures (such as MIPS, and some
+ * SPARC) .. is this possible for SuperH? (This is a non-issue if the
+ * SH4 cache is configured in write-through mode).
+ *
+ * In the meantime, we'll just flush all of the caches if we have a
+ * valid mm context.. this seems to be the simplest way to avoid at
+ * least a few wasted cache flushes. -Lethal
+ */
+ if (mm->context !=3D 0)
+ flush_cache_all();
}
=20
/*
@@ -298,14 +326,39 @@
void flush_cache_range(struct mm_struct *mm, unsigned long start,
unsigned long end)
{
- /*
- * We could call flush_cache_page for the pages of these range,
- * but it's not efficient (scan the caches all the time...).
- *
- * We can't use A-bit magic, as there's the case we don't have
- * valid entry on TLB.
- */
- flush_cache_all();
+ unsigned long flags;
+
+ if (mm->context =3D=3D 0)
+ return;
+
+ start &=3D PAGE_MASK;
+
+ if (!find_vma(mm, start))
+ return;
+ if (mm->context !=3D current->active_mm->context) {
+ flush_cache_all();
+ } else {
+ pgd_t *pgd;
+ pmd_t *pmd;
+ pte_t *pte;
+
+ save_and_cli(flags);
+ jump_to_P2();
+
+ for (start; start < end; start +=3D PAGE_SIZE) {
+ pgd =3D pgd_offset(mm, start);
+ pmd =3D pmd_offset(pgd, start);
+ pte =3D pte_offset(pmd, start);
+
+ if (pte_val(*pte) & _PAGE_PRESENT) {
+ __flush_icache_page(start);
+ __flush_dcache_page(start);
+ }
+ }
+
+ back_to_P1();
+ restore_flags(flags);
+ }
}
=20
/*
|
