| Message ID | 20220920151202.180057-5-chenzhongjin@huawei.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | riscv: Improvments for stacktrace | expand |
On Tue, Sep 20, 2022 at 11:15 PM Chen Zhongjin <chenzhongjin@huawei.com> wrote: > > syscall uses s0 to load address of sys_call_table. > > Since now we uses s0 to save pt_regs for unwinding, clobber > s0 can make unwinder treat s0 as pt_regs address. Use s1 for > this job. This patch does not make sense. Why couldn't we use s1 for pt_regs? No need to modify the entry.S here. > > Signed-off-by: Chen Zhongjin <chenzhongjin@huawei.com> > --- > arch/riscv/kernel/entry.S | 20 ++++++++++---------- > 1 file changed, 10 insertions(+), 10 deletions(-) > > diff --git a/arch/riscv/kernel/entry.S b/arch/riscv/kernel/entry.S > index ecb15c7430b4..a3b14a649782 100644 > --- a/arch/riscv/kernel/entry.S > +++ b/arch/riscv/kernel/entry.S > @@ -205,33 +205,33 @@ handle_syscall: > check_syscall_nr: > /* Check to make sure we don't jump to a bogus syscall number. */ > li t0, __NR_syscalls > - la s0, sys_ni_syscall > + la s1, sys_ni_syscall > /* > * Syscall number held in a7. > * If syscall number is above allowed value, redirect to ni_syscall. > */ > bgeu a7, t0, 3f > #ifdef CONFIG_COMPAT > - REG_L s0, PT_STATUS(sp) > - srli s0, s0, SR_UXL_SHIFT > - andi s0, s0, (SR_UXL >> SR_UXL_SHIFT) > + REG_L s1, PT_STATUS(sp) > + srli s1, s1, SR_UXL_SHIFT > + andi s1, s1, (SR_UXL >> SR_UXL_SHIFT) > li t0, (SR_UXL_32 >> SR_UXL_SHIFT) > - sub t0, s0, t0 > + sub t0, s1, t0 > bnez t0, 1f > > /* Call compat_syscall */ > - la s0, compat_sys_call_table > + la s1, compat_sys_call_table > j 2f > 1: > #endif > /* Call syscall */ > - la s0, sys_call_table > + la s1, sys_call_table > 2: > slli t0, a7, RISCV_LGPTR > - add s0, s0, t0 > - REG_L s0, 0(s0) > + add s1, s1, t0 > + REG_L s1, 0(s1) > 3: > - jalr s0 > + jalr s1 > > ret_from_syscall: > /* Set user a0 to kernel a0 */ > -- > 2.17.1 >
On Wed, Sep 21, 2022 at 9:30 AM Guo Ren <guoren@kernel.org> wrote: > > On Tue, Sep 20, 2022 at 11:15 PM Chen Zhongjin <chenzhongjin@huawei.com> wrote: > > > > syscall uses s0 to load address of sys_call_table. > > > > Since now we uses s0 to save pt_regs for unwinding, clobber > > s0 can make unwinder treat s0 as pt_regs address. Use s1 for > > this job. > This patch does not make sense. Why couldn't we use s1 for pt_regs? Seems s0->fp can't be replaced. I take back the question. > > No need to modify the entry.S here. > > > > > Signed-off-by: Chen Zhongjin <chenzhongjin@huawei.com> > > --- > > arch/riscv/kernel/entry.S | 20 ++++++++++---------- > > 1 file changed, 10 insertions(+), 10 deletions(-) > > > > diff --git a/arch/riscv/kernel/entry.S b/arch/riscv/kernel/entry.S > > index ecb15c7430b4..a3b14a649782 100644 > > --- a/arch/riscv/kernel/entry.S > > +++ b/arch/riscv/kernel/entry.S > > @@ -205,33 +205,33 @@ handle_syscall: > > check_syscall_nr: > > /* Check to make sure we don't jump to a bogus syscall number. */ > > li t0, __NR_syscalls > > - la s0, sys_ni_syscall > > + la s1, sys_ni_syscall > > /* > > * Syscall number held in a7. > > * If syscall number is above allowed value, redirect to ni_syscall. > > */ > > bgeu a7, t0, 3f > > #ifdef CONFIG_COMPAT > > - REG_L s0, PT_STATUS(sp) > > - srli s0, s0, SR_UXL_SHIFT > > - andi s0, s0, (SR_UXL >> SR_UXL_SHIFT) > > + REG_L s1, PT_STATUS(sp) > > + srli s1, s1, SR_UXL_SHIFT > > + andi s1, s1, (SR_UXL >> SR_UXL_SHIFT) > > li t0, (SR_UXL_32 >> SR_UXL_SHIFT) > > - sub t0, s0, t0 > > + sub t0, s1, t0 > > bnez t0, 1f > > > > /* Call compat_syscall */ > > - la s0, compat_sys_call_table > > + la s1, compat_sys_call_table > > j 2f > > 1: > > #endif > > /* Call syscall */ > > - la s0, sys_call_table > > + la s1, sys_call_table > > 2: > > slli t0, a7, RISCV_LGPTR > > - add s0, s0, t0 > > - REG_L s0, 0(s0) > > + add s1, s1, t0 > > + REG_L s1, 0(s1) > > 3: > > - jalr s0 > > + jalr s1 > > > > ret_from_syscall: > > /* Set user a0 to kernel a0 */ > > -- > > 2.17.1 > > > > > -- > Best Regards > Guo Ren
diff --git a/arch/riscv/kernel/entry.S b/arch/riscv/kernel/entry.S index ecb15c7430b4..a3b14a649782 100644 --- a/arch/riscv/kernel/entry.S +++ b/arch/riscv/kernel/entry.S @@ -205,33 +205,33 @@ handle_syscall: check_syscall_nr: /* Check to make sure we don't jump to a bogus syscall number. */ li t0, __NR_syscalls - la s0, sys_ni_syscall + la s1, sys_ni_syscall /* * Syscall number held in a7. * If syscall number is above allowed value, redirect to ni_syscall. */ bgeu a7, t0, 3f #ifdef CONFIG_COMPAT - REG_L s0, PT_STATUS(sp) - srli s0, s0, SR_UXL_SHIFT - andi s0, s0, (SR_UXL >> SR_UXL_SHIFT) + REG_L s1, PT_STATUS(sp) + srli s1, s1, SR_UXL_SHIFT + andi s1, s1, (SR_UXL >> SR_UXL_SHIFT) li t0, (SR_UXL_32 >> SR_UXL_SHIFT) - sub t0, s0, t0 + sub t0, s1, t0 bnez t0, 1f /* Call compat_syscall */ - la s0, compat_sys_call_table + la s1, compat_sys_call_table j 2f 1: #endif /* Call syscall */ - la s0, sys_call_table + la s1, sys_call_table 2: slli t0, a7, RISCV_LGPTR - add s0, s0, t0 - REG_L s0, 0(s0) + add s1, s1, t0 + REG_L s1, 0(s1) 3: - jalr s0 + jalr s1 ret_from_syscall: /* Set user a0 to kernel a0 */
syscall uses s0 to load address of sys_call_table. Since now we uses s0 to save pt_regs for unwinding, clobber s0 can make unwinder treat s0 as pt_regs address. Use s1 for this job. Signed-off-by: Chen Zhongjin <chenzhongjin@huawei.com> --- arch/riscv/kernel/entry.S | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-)