* [PATCH v2] loongarch: kgdb: remove unnecessary init variable
@ 2025-06-04 6:56 Li Jun
2025-06-04 14:11 ` Huacai Chen
0 siblings, 1 reply; 6+ messages in thread
From: Li Jun @ 2025-06-04 6:56 UTC (permalink / raw)
To: si.yanteng, chenhuacai, kernel, loongarch, lijun01
The compiler generates initialization instructions,
which consume additional CPU cycles. the
get_step_address(regs, &addr) should assign a value
to 'error' before it is read.so the var don't need init
to 0.
Signed-off-by: Li Jun <lijun01@kylinos.cn>
---
arch/loongarch/kernel/kgdb.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/arch/loongarch/kernel/kgdb.c b/arch/loongarch/kernel/kgdb.c
index 7be5b4c0c900..df85d6a0dc74 100644
--- a/arch/loongarch/kernel/kgdb.c
+++ b/arch/loongarch/kernel/kgdb.c
@@ -379,10 +379,9 @@ static int get_step_address(struct pt_regs *regs, unsigned long *next_addr)
static int do_single_step(struct pt_regs *regs)
{
- int error = 0;
unsigned long addr = 0; /* Determine where the target instruction will send us to */
+ int error = get_step_address(regs, &addr);
- error = get_step_address(regs, &addr);
if (error)
return error;
--
2.25.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH v2] loongarch: kgdb: remove unnecessary init variable
2025-06-04 6:56 [PATCH v2] loongarch: kgdb: remove unnecessary init variable Li Jun
@ 2025-06-04 14:11 ` Huacai Chen
2025-06-05 1:30 ` Re：[PATCH " lijun
2025-06-05 1:41 ` Re：[PATCH " lijun
0 siblings, 2 replies; 6+ messages in thread
From: Huacai Chen @ 2025-06-04 14:11 UTC (permalink / raw)
To: Li Jun; +Cc: si.yanteng, kernel, loongarch
On Wed, Jun 4, 2025 at 2:57 PM Li Jun <lijun01@kylinos.cn> wrote:
>
> The compiler generates initialization instructions,
> which consume additional CPU cycles. the
> get_step_address(regs, &addr) should assign a value
> to 'error' before it is read.so the var don't need init
> to 0.
Why do you think the compiler generates initialization instructions,
have you found any difference in the binary after your patch?
Huacai
>
> Signed-off-by: Li Jun <lijun01@kylinos.cn>
> ---
> arch/loongarch/kernel/kgdb.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/arch/loongarch/kernel/kgdb.c b/arch/loongarch/kernel/kgdb.c
> index 7be5b4c0c900..df85d6a0dc74 100644
> --- a/arch/loongarch/kernel/kgdb.c
> +++ b/arch/loongarch/kernel/kgdb.c
> @@ -379,10 +379,9 @@ static int get_step_address(struct pt_regs *regs, unsigned long *next_addr)
>
> static int do_single_step(struct pt_regs *regs)
> {
> - int error = 0;
> unsigned long addr = 0; /* Determine where the target instruction will send us to */
> + int error = get_step_address(regs, &addr);
>
> - error = get_step_address(regs, &addr);
> if (error)
> return error;
>
> --
> 2.25.1
>
>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re：[PATCH v2] loongarch: kgdb: remove unnecessary init variable
2025-06-04 14:11 ` Huacai Chen
@ 2025-06-05 1:30 ` lijun
2025-06-05 1:37 ` [PATCH " Huacai Chen
2025-06-05 1:41 ` Re：[PATCH " lijun
1 sibling, 1 reply; 6+ messages in thread
From: lijun @ 2025-06-05 1:30 UTC (permalink / raw)
To: lijun01; +Cc: kernel, loongarch
#include<stdio.h> #include
<stdio.h>
int main(int argc, const char *argv[]) int main(int argc, const
char *argv[])
{
{
int
i=0;
int i;
i =
3;
i = 3;
printf("i = %d\n",
i); printf("i = %d\n", i);
return
0; return
0;
}
}
endbr64
endbr64
push %rbp p
ush %rbp
mov %rsp,%rbp mov %r
sp,%rbp
sub $0x20,%rsp sub
$0x20,%rsp
mov %edi,-0x14(%rbp) mov %edi,-
0x14(%rbp)
mov %rsi,-0x20(%rbp) mov %rsi,-
0x20(%rbp)
movl $0x0,-0x4(%rbp) movl $0x3,-
0x4(%rbp)
movl $0x3,-0x4(%rbp) mov -
0x4(%rbp),%eax
mov -0x4(%rbp),%eax mov %eax,%esi
mov %eax,%esi mov $0
x402004,%edi
mov $0x402004,%edi mov $0x0,%eax
mov $0x0,%eax callq 4
01040 <printf@plt>
callq 401040 <printf@plt>
if gcc -O3 test2.c -o a2.out, It is possible that the
initialization is optimized by
assigning values immediately after initialization
endbr64
sub $0x8,%rsp
mov $0x3,%edx
mov $0x402004,%esi
xor %eax,%eax
mov $0x1,%edi
callq 401040 <__printf_chk@plt>
----------------------------------------------->
Why do you think the compiler generates initialization instructions,
have you found any difference in the binary after your patch?
Huacai
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2] loongarch: kgdb: remove unnecessary init variable
2025-06-05 1:30 ` Re：[PATCH " lijun
@ 2025-06-05 1:37 ` Huacai Chen
0 siblings, 0 replies; 6+ messages in thread
From: Huacai Chen @ 2025-06-05 1:37 UTC (permalink / raw)
To: lijun; +Cc: kernel, loongarch
On Thu, Jun 5, 2025 at 9:31 AM lijun <lijun01@kylinos.cn> wrote:
>
>
>
> #include<stdio.h> #include
> <stdio.h>
>
> int main(int argc, const char *argv[]) int main(int argc, const
> char *argv[])
> {
> {
> int
> i=0;
> int i;
>
>
> i =
> 3;
> i = 3;
> printf("i = %d\n",
> i); printf("i = %d\n", i);
> return
> 0; return
> 0;
> }
> }
> endbr64
> endbr64
> push %rbp p
> ush %rbp
> mov %rsp,%rbp mov %r
> sp,%rbp
> sub $0x20,%rsp sub
> $0x20,%rsp
> mov %edi,-0x14(%rbp) mov %edi,-
> 0x14(%rbp)
> mov %rsi,-0x20(%rbp) mov %rsi,-
> 0x20(%rbp)
> movl $0x0,-0x4(%rbp) movl $0x3,-
> 0x4(%rbp)
> movl $0x3,-0x4(%rbp) mov -
> 0x4(%rbp),%eax
> mov -0x4(%rbp),%eax mov %eax,%esi
> mov %eax,%esi mov $0
> x402004,%edi
> mov $0x402004,%edi mov $0x0,%eax
> mov $0x0,%eax callq 4
> 01040 <printf@plt>
> callq 401040 <printf@plt>
>
> if gcc -O3 test2.c -o a2.out, It is possible that the
> initialization is optimized by
> assigning values immediately after initialization
> endbr64
> sub $0x8,%rsp
> mov $0x3,%edx
> mov $0x402004,%esi
> xor %eax,%eax
> mov $0x1,%edi
> callq 401040 <__printf_chk@plt>
The format is completely unreadable here:
https://lore.kernel.org/loongarch/CAAhV-H5dmh8_rNNRAaFeFvjCtt0CCsqTxOujWFakU=RYakrpAA@mail.gmail.com/T/#t
And we are talking about the kernel of LoongArch, you provide a user
program of x86?
Huacai
>
> ----------------------------------------------->
> Why do you think the compiler generates initialization instructions,
> have you found any difference in the binary after your patch?
>
> Huacai
>
>
>
>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re：[PATCH v2] loongarch: kgdb: remove unnecessary init variable
2025-06-04 14:11 ` Huacai Chen
2025-06-05 1:30 ` Re：[PATCH " lijun
@ 2025-06-05 1:41 ` lijun
2025-06-05 1:46 ` [PATCH " Huacai Chen
1 sibling, 1 reply; 6+ messages in thread
From: lijun @ 2025-06-05 1:41 UTC (permalink / raw)
To: Huacai Chen; +Cc: si.yanteng, kernel, loongarch
//test2.c
#include<stdio.h>
int main(int argc, const char *argv[])
{
int
i=0;
i = 3;
printf("i = %d\n", i);
return 0;
}
push %rbp
mov %rsp,%rbp
sub $0x20,%rsp
mov %edi,-0x14(%rbp)
mov %rsi,-0x20(%rbp)
movl $0x0,-0x4(%rbp)
movl $0x3,-0x4(%rbp)
mov -0x4(%rbp),%eax
mov %eax,%esi
mov $0x402004,%edi
mov $0x0,%eax
callq 401040 <printf@plt>
//test.c
#include<stdio.h>
int main(int argc, const char *argv[])
{
int i;
i = 3;
printf("i = %d\n", i);
return 0;
}
endbr64
push %rbp
mov %rsp,%rbp
sub $0x20,%rsp
mov %edi,-0x14(%rbp)
mov %rsi,-0x20(%rbp)
movl $0x3,-0x4(%rbp)
mov -0x4(%rbp),%eax
mov %eax,%esi
mov $0x402004,%edi
mov $0x0,%eax
callq 401040 <printf@plt>
It is possible that the initialization is optimized by assigning values
immediately after initialization
if gcc -O3 test2.c -o ao3.out
endbr64
sub $0x8,%rsp
mov $0x3,%edx
mov $0x402004,%esi
xor %eax,%eax
mov $0x1,%edi
callq 401040 <__printf_chk@plt>
----------------------------------->
Why do you think the compiler generates initialization instructions,
have you found any difference in the binary after your patch?
Huacai
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2] loongarch: kgdb: remove unnecessary init variable
2025-06-05 1:41 ` Re：[PATCH " lijun
@ 2025-06-05 1:46 ` Huacai Chen
0 siblings, 0 replies; 6+ messages in thread
From: Huacai Chen @ 2025-06-05 1:46 UTC (permalink / raw)
To: lijun; +Cc: si.yanteng, kernel, loongarch
On Thu, Jun 5, 2025 at 9:41 AM lijun <lijun01@kylinos.cn> wrote:
>
>
> //test2.c
> #include<stdio.h>
>
> int main(int argc, const char *argv[])
> {
> int
> i=0;
> i = 3;
> printf("i = %d\n", i);
> return 0;
> }
> push %rbp
> mov %rsp,%rbp
> sub $0x20,%rsp
> mov %edi,-0x14(%rbp)
> mov %rsi,-0x20(%rbp)
> movl $0x0,-0x4(%rbp)
> movl $0x3,-0x4(%rbp)
> mov -0x4(%rbp),%eax
> mov %eax,%esi
> mov $0x402004,%edi
> mov $0x0,%eax
> callq 401040 <printf@plt>
> //test.c
> #include<stdio.h>
>
> int main(int argc, const char *argv[])
> {
> int i;
> i = 3;
> printf("i = %d\n", i);
> return 0;
> }
> endbr64
> push %rbp
> mov %rsp,%rbp
> sub $0x20,%rsp
> mov %edi,-0x14(%rbp)
> mov %rsi,-0x20(%rbp)
> movl $0x3,-0x4(%rbp)
> mov -0x4(%rbp),%eax
> mov %eax,%esi
> mov $0x402004,%edi
> mov $0x0,%eax
> callq 401040 <printf@plt>
>
> It is possible that the initialization is optimized by assigning values
> immediately after initialization
> if gcc -O3 test2.c -o ao3.out
> endbr64
> sub $0x8,%rsp
> mov $0x3,%edx
> mov $0x402004,%esi
> xor %eax,%eax
> mov $0x1,%edi
> callq 401040 <__printf_chk@plt>
Still a user program of x86, not the kernel of LoongArch.
And top posting is strictly forbidden in the kernel community which
Yanteng has already pointed out. Don't do it again and again, and
don't always only answer part of reviewers comments.
Huacai
>
>
>
> ----------------------------------->
> Why do you think the compiler generates initialization instructions,
> have you found any difference in the binary after your patch?
>
> Huacai
>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2025-06-05 1:46 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-06-04 6:56 [PATCH v2] loongarch: kgdb: remove unnecessary init variable Li Jun
2025-06-04 14:11 ` Huacai Chen
2025-06-05 1:30 ` Re：[PATCH " lijun
2025-06-05 1:37 ` [PATCH " Huacai Chen
2025-06-05 1:41 ` Re：[PATCH " lijun
2025-06-05 1:46 ` [PATCH " Huacai Chen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).