| Message ID | 20241025100700.3714552-5-ruanjinjie@huawei.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | arm64: entry: Convert to generic entry | expand |
On Fri, Oct 25, 2024 at 06:06:45PM +0800, Jinjie Ruan wrote: > The __enter_from_kernel_mode() is only called by enter_from_kernel_mode(), > remove it. The point of this split is to cleanly separate the raw entry logic (in __enter_from_kernel_mode() from pieces that run later and can safely be instrumented (later in enter_from_kernel_mode()). I had expected that a later patch would replace __enter_from_kernel_mode() with the generic equivalent, leaving enter_from_kernel_mode() unchanged. It looks like patch 16 could do that without this patch being necessary -- am I missing something? Mark. > > No functional changes. > > Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com> > --- > arch/arm64/kernel/entry-common.c | 9 +-------- > 1 file changed, 1 insertion(+), 8 deletions(-) > > diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c > index ccf59b44464d..a7fd4d6c7650 100644 > --- a/arch/arm64/kernel/entry-common.c > +++ b/arch/arm64/kernel/entry-common.c > @@ -36,7 +36,7 @@ > * This is intended to match the logic in irqentry_enter(), handling the kernel > * mode transitions only. > */ > -static __always_inline irqentry_state_t __enter_from_kernel_mode(struct pt_regs *regs) > +static noinstr irqentry_state_t enter_from_kernel_mode(struct pt_regs *regs) > { > irqentry_state_t ret = { > .exit_rcu = false, > @@ -55,13 +55,6 @@ static __always_inline irqentry_state_t __enter_from_kernel_mode(struct pt_regs > rcu_irq_enter_check_tick(); > trace_hardirqs_off_finish(); > > - return ret; > -} > - > -static noinstr irqentry_state_t enter_from_kernel_mode(struct pt_regs *regs) > -{ > - irqentry_state_t ret = __enter_from_kernel_mode(regs); > - > mte_check_tfsr_entry(); > mte_disable_tco_entry(current); > > -- > 2.34.1 >
On 2024/10/29 22:37, Mark Rutland wrote: > On Fri, Oct 25, 2024 at 06:06:45PM +0800, Jinjie Ruan wrote: >> The __enter_from_kernel_mode() is only called by enter_from_kernel_mode(), >> remove it. > > The point of this split is to cleanly separate the raw entry logic (in > __enter_from_kernel_mode() from pieces that run later and can safely be > instrumented (later in enter_from_kernel_mode()). Hi, Mark, I reviewed your commit bc29b71f53b1 ("arm64: entry: clarify entry/exit helpers"), and keep these functions is to make instrumentation boundaries more clear, and will not change them. > > I had expected that a later patch would replace > __enter_from_kernel_mode() with the generic equivalent, leaving > enter_from_kernel_mode() unchanged. It looks like patch 16 could do that > without this patch being necessary -- am I missing something? Yes, you are right! these useless cleanup patches will be removed. And when switched to generic syscall, I found that proper refactoring would also facilitate clear code switching. Thank you. > > Mark. > >> >> No functional changes. >> >> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com> >> --- >> arch/arm64/kernel/entry-common.c | 9 +-------- >> 1 file changed, 1 insertion(+), 8 deletions(-) >> >> diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c >> index ccf59b44464d..a7fd4d6c7650 100644 >> --- a/arch/arm64/kernel/entry-common.c >> +++ b/arch/arm64/kernel/entry-common.c >> @@ -36,7 +36,7 @@ >> * This is intended to match the logic in irqentry_enter(), handling the kernel >> * mode transitions only. >> */ >> -static __always_inline irqentry_state_t __enter_from_kernel_mode(struct pt_regs *regs) >> +static noinstr irqentry_state_t enter_from_kernel_mode(struct pt_regs *regs) >> { >> irqentry_state_t ret = { >> .exit_rcu = false, >> @@ -55,13 +55,6 @@ static __always_inline irqentry_state_t __enter_from_kernel_mode(struct pt_regs >> rcu_irq_enter_check_tick(); >> trace_hardirqs_off_finish(); >> >> - return ret; >> -} >> - >> -static noinstr irqentry_state_t enter_from_kernel_mode(struct pt_regs *regs) >> -{ >> - irqentry_state_t ret = __enter_from_kernel_mode(regs); >> - >> mte_check_tfsr_entry(); >> mte_disable_tco_entry(current); >> >> -- >> 2.34.1 >> >
diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c index ccf59b44464d..a7fd4d6c7650 100644 --- a/arch/arm64/kernel/entry-common.c +++ b/arch/arm64/kernel/entry-common.c @@ -36,7 +36,7 @@ * This is intended to match the logic in irqentry_enter(), handling the kernel * mode transitions only. */ -static __always_inline irqentry_state_t __enter_from_kernel_mode(struct pt_regs *regs) +static noinstr irqentry_state_t enter_from_kernel_mode(struct pt_regs *regs) { irqentry_state_t ret = { .exit_rcu = false, @@ -55,13 +55,6 @@ static __always_inline irqentry_state_t __enter_from_kernel_mode(struct pt_regs rcu_irq_enter_check_tick(); trace_hardirqs_off_finish(); - return ret; -} - -static noinstr irqentry_state_t enter_from_kernel_mode(struct pt_regs *regs) -{ - irqentry_state_t ret = __enter_from_kernel_mode(regs); - mte_check_tfsr_entry(); mte_disable_tco_entry(current);
The __enter_from_kernel_mode() is only called by enter_from_kernel_mode(), remove it. No functional changes. Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com> --- arch/arm64/kernel/entry-common.c | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-)