| Message ID | 1427358533-3754-4-git-send-email-takahiro.akashi@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, 2015-03-26 at 17:28 +0900, AKASHI Takahiro wrote: > @@ -64,7 +65,11 @@ void soft_restart(unsigned long addr) > setup_mm_for_reboot(); > > cpu_soft_restart(virt_to_phys(cpu_reset), > - is_hyp_mode_available(), addr); > +#ifdef CONFIG_KEXEC > + !in_crash_kexec && > +#endif Why not define in_crash_kexec without condition on CONFIG_KEXEC, say here in process.c and then avoid these preprocessor conditionals. > + is_hyp_mode_available(), > + addr); > > /* Should never get here */ > BUG();
On 03/27/2015 07:29 AM, Geoff Levand wrote: > On Thu, 2015-03-26 at 17:28 +0900, AKASHI Takahiro wrote: >> @@ -64,7 +65,11 @@ void soft_restart(unsigned long addr) >> setup_mm_for_reboot(); >> >> cpu_soft_restart(virt_to_phys(cpu_reset), >> - is_hyp_mode_available(), addr); >> +#ifdef CONFIG_KEXEC >> + !in_crash_kexec && >> +#endif > > Why not define in_crash_kexec without condition on CONFIG_KEXEC, say > here in process.c and then avoid these preprocessor conditionals. Well, I thought of that, but as its name suggested, the variable should be basically part of kdump code. Moreover, I suspect that some one might reject a whole idea of patch #3. -Takahiro AKASHI >> + is_hyp_mode_available(), >> + addr); >> >> /* Should never get here */ >> BUG(); > > >
diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c index d894d3e..9859f5c 100644 --- a/arch/arm64/kernel/process.c +++ b/arch/arm64/kernel/process.c @@ -48,6 +48,7 @@ #include <asm/compat.h> #include <asm/cacheflush.h> #include <asm/fpsimd.h> +#include <asm/kexec.h> #include <asm/mmu_context.h> #include <asm/processor.h> #include <asm/stacktrace.h> @@ -64,7 +65,11 @@ void soft_restart(unsigned long addr) setup_mm_for_reboot(); cpu_soft_restart(virt_to_phys(cpu_reset), - is_hyp_mode_available(), addr); +#ifdef CONFIG_KEXEC + !in_crash_kexec && +#endif + is_hyp_mode_available(), + addr); /* Should never get here */ BUG();
Unlike normal kexec case, we don't have a chance to reset EL2 context in a generic way because bad exceptions may directly invoke crash_kexec(). (See die().) Kvm is not useful on crash dump kernel anyway, and so we let it un-initialized across rebooting. Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org> --- arch/arm64/kernel/process.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)