Techwizz/kernel-fxtec-pro1x
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Blackfin: optimize fixed code handling for the most common case

Browse source 
The majority of the time we are returning to user space, it is not in the
fixed atomic code region.  So rather than branch to a function where we
check the PC and return, do the check inline and branch only when needed.

Also, tweak some of the fixed code handling based on assumptions we are
aware of but cannot be expressed in C.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
This commit is contained in:
Mike Frysinger 2009-08-17 19:05:07 +00:00
parent dac981740f
commit 926494943b
2 changed files with 8 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
arch/blackfin/kernel/process.c
View file

@ -282,25 +282,19 @@ void finish_atomic_sections (struct pt_regs *regs)
{
int __user *up0 = (int __user *)regs->p0;
if (regs->pc < ATOMIC_SEQS_START || regs->pc >= ATOMIC_SEQS_END)
return;
switch (regs->pc) {
case ATOMIC_XCHG32 + 2:
put_user(regs->r1, up0);
regs->pc += 2;
regs->pc = ATOMIC_XCHG32 + 4;
break;
case ATOMIC_CAS32 + 2:
case ATOMIC_CAS32 + 4:
if (regs->r0 == regs->r1)
case ATOMIC_CAS32 + 6:
put_user(regs->r2, up0);
regs->pc = ATOMIC_CAS32 + 8;
break;
case ATOMIC_CAS32 + 6:
put_user(regs->r2, up0);
regs->pc += 2;
break;
case ATOMIC_ADD32 + 2:
regs->r0 = regs->r1 + regs->r0;

6
arch/blackfin/mach-common/entry.S
View file

@ -1025,6 +1025,12 @@ ENTRY(_schedule_and_signal_from_int)
#endif
sti r0;
/* finish the userspace "atomic" functions for it */
r1 = FIXED_CODE_END;
r2 = [sp + PT_PC];
cc = r1 <= r2;
if cc jump .Lresume_userspace (bp);
r0 = sp;
sp += -12;
call _finish_atomic_sections;