[PATCH 13/23] arm64: introduce is_a32_task and is_a32_thread (for AArch32 compat)

Yury Norov ynorov at caviumnetworks.com
Sun Jun 12 10:56:44 PDT 2016


On Sun, Jun 12, 2016 at 09:08:28PM +0800, Zhangjian (Bamvor) wrote:
> >>  #ifdef CONFIG_COMPAT
> >>  #define TASK_SIZE_32        UL(0x100000000)
> >>-#define TASK_SIZE        (test_thread_flag(TIF_32BIT) ? \
> >>+#define TASK_SIZE        (is_compat_task() ?        \
> >>                  TASK_SIZE_32 : TASK_SIZE_64)
> >>  #define TASK_SIZE_OF(tsk)    (test_tsk_thread_flag(tsk, TIF_32BIT) ? \
> >>                  TASK_SIZE_32 : TASK_SIZE_64)
> >Should we update or delete this macro?
> >  #define TASK_SIZE_OF(tsk) (is_compat_task() ? \
> >TASK_SIZE_32 : TASK_SIZE_64)
> Sorry it should be:
> #define TASK_SIZE_OF(tsk)       ((is_a32_compat_thread(task_thread_info(tsk)) \
>                                   || is_ilp32_compat_thread(task_thread_info(tsk))) ? \
>                                 TASK_SIZE_32 : TASK_SIZE_64)

Thank you. I know about this, but always forget to fix. )
I think we'd introduce is_compat_thread() as well.

> >x86, sparc, mips, ppc, parisc, s390 define its own version. But
> >"include/linux/sched.h" will define it if
> >TASK_SIZE_OF does not exist:
> >#ifndef TASK_SIZE_OF
> >#define TASK_SIZE_OF(tsk)       TASK_SIZE
> >#endif
> >
> >
> >Regards
> >
> >Bamvor
> >



More information about the linux-arm-kernel mailing list