#5217 closed enhancement (fixed)
[PATCH] speed-up x86_enter_userspace
| Reported by: | Grey | Owned by: | axeld |
|---|---|---|---|
| Priority: | normal | Milestone: | R1 |
| Component: | System/Kernel | Version: | R1/Development |
| Keywords: | arch_x86.S | Cc: | |
| Blocked By: | Blocking: | ||
| Platform: | x86 |
Description
Small optimization of x86_enter_userspace. Patch attached.
Attachments (1)
Change History (8)
by , 14 years ago
| Attachment: | x86_enter_userspace.patch added |
|---|
follow-up: 3 comment:1 by , 14 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
follow-up: 4 comment:3 by , 14 years ago
Replying to bonefish:
I didn't see the pointer for moving the clearing of %ebp, so I left it out, but the rest I applied in hrev34949.
Thanks a lot! BTW, how about removing i386_switch_stack_and_call and arch_thread_switch_kstack_and_call from src/system/kernel/arch/x86/arch_thread.cpp ? And similar for all other cpu. It's not used at all.
follow-up: 5 comment:4 by , 14 years ago
follow-up: 6 comment:5 by , 14 years ago
Replying to bonefish:
How about removing memory_read_barrier(void) from src/system/kernel/smp.cpp and arch_cpu_memory_read_barrier for all cpu. It's not used at all.
follow-up: 7 comment:6 by , 14 years ago
Replying to Grey:
Replying to bonefish:
How about removing memory_read_barrier(void) from src/system/kernel/smp.cpp and arch_cpu_memory_read_barrier for all cpu. It's not used at all.
Yeah, but it's generally useful functionality and we might even want to make it public one day.
BTW, the haiku-development or the kernel development mailing lists are much more suitable for this kind of discussion.
comment:7 by , 14 years ago
Replying to bonefish:
Replying to Grey:
Replying to bonefish:
How about removing memory_read_barrier(void) from src/system/kernel/smp.cpp and arch_cpu_memory_read_barrier for all cpu. It's not used at all.
Yeah, but it's generally useful functionality and we might even want to make it public one day.
Ok. Hint: Let's do this functions inline. Unused inline functions will be absent in obj flie.
BTW, the haiku-development or the kernel development mailing lists are much more suitable for this kind of discussion.
See you in mailing list.
I didn't see the pointer for moving the clearing of %ebp, so I left it out, but the rest I applied in hrev34949.