Re: [Qemu-devel] [PATCH 03/13] target: arm: Move all v7m helpers into their own file

[Peter Maydell]

On 13 November 2018 at 16:52, Samuel Ortiz <address@hidden> wrote:
> In preparation for supporting TCG disablement on ARM, we move all TCG
> related v7m helpers and APIs into their own file (m_helper.c for all
> v*-m helpers).
> arm_v7m_cpu_do_interrupt pulls a large number of static functions
> out of helper.c into m_helper.c because it is TCG dependent.
>
> Signed-off-by: Samuel Ortiz <address@hidden>
> Signed-off-by: Philippe Mathieu-Daudé <address@hidden>
> Tested-by: Philippe Mathieu-Daudé <address@hidden>
> Reviewed-by: Robert Bradford <address@hidden>
> ---
>  target/arm/internals.h   |   37 +
>  target/arm/helper.c      | 2209 +++-----------------------------------
>  target/arm/m_helper.c    | 1892 ++++++++++++++++++++++++++++++++
>  target/arm/Makefile.objs |    2 +-
>  4 files changed, 2086 insertions(+), 2054 deletions(-)
>  create mode 100644 target/arm/m_helper.c

> +/* Function used to synchronize QEMU's AArch64 register set with AArch32
> + * register set.  This is necessary when switching between AArch32 and 
> AArch64
> + * execution state.
> + */
> +void aarch64_sync_32_to_64(CPUARMState *env)
>  {
> -    uint32_t new_ss_msp, new_ss_psp;
> +    int i;
> +    uint32_t mode = env->uncached_cpsr & CPSR_M;
>
> -    if (env->v7m.secure == new_secstate) {
> -        return;
> +    /* We can blanket copy R[0:7] to X[0:7] */
> +    for (i = 0; i < 8; i++) {
> +        env->xregs[i] = env->regs[i];
>      }
>
> -    /* All the banked state is accessed by looking at env->v7m.secure
> -     * except for the stack pointer; rearrange the SP appropriately.
> +    /* Unless we are in FIQ mode, x8-x12 come from the user registers r8-r12.
> +     * Otherwise, they come from the banked user regs.
>       */
> -    new_ss_msp = env->v7m.other_ss_msp;
> -    new_ss_psp = env->v7m.other_ss_psp;
> -
> -    if (v7m_using_psp(env)) {
> -        env->v7m.other_ss_psp = env->regs[13];
> -        env->v7m.other_ss_msp = env->v7m.other_sp;
> -    } else {
> -        env->v7m.other_ss_msp = env->regs[13];
> -        env->v7m.other_ss_psp = env->v7m.other_sp;
> -    }
> -
> -    env->v7m.secure = new_secstate;
> -
> -    if (v7m_using_psp(env)) {
> -        env->regs[13] = new_ss_psp;
> -        env->v7m.other_sp = new_ss_msp;
> +    if (mode == ARM_CPU_MODE_FIQ) {
> +        for (i = 8; i < 13; i++) {
> +            env->xregs[i] = env->usr_regs[i - 8];
> +        }
>      } else {
> -        env->regs[13] = new_ss_msp;
> -        env->v7m.other_sp = new_ss_psp;
> +        for (i = 8; i < 13; i++) {
> +            env->xregs[i] = env->regs[i];
> +        }
>      }
> -}
>
> -void HELPER(v7m_bxns)(CPUARMState *env, uint32_t dest)
> -{
> -    /* Handle v7M BXNS:
> -     *  - if the return value is a magic value, do exception return (like BX)
> -     *  - otherwise bit 0 of the return value is the target security state
> +    /* Registers x13-x23 are the various mode SP and FP registers. Registers
> +     * r13 and r14 are only copied if we are in that mode, otherwise we copy
> +     * from the mode banked register.
>       */
> -    uint32_t min_magic;
> -
> -    if (arm_feature(env, ARM_FEATURE_M_SECURITY)) {
> -        /* Covers FNC_RETURN and EXC_RETURN magic */
> -        min_magic = FNC_RETURN_MIN_MAGIC;
> +    if (mode == ARM_CPU_MODE_USR || mode == ARM_CPU_MODE_SYS) {
> +        env->xregs[13] = env->regs[13];
> +        env->xregs[14] = env->regs[14];
>      } else {
> -        /* EXC_RETURN magic only */
> -        min_magic = EXC_RETURN_MIN_MAGIC;
> +        env->xregs[13] = env->banked_r13[bank_number(ARM_CPU_MODE_USR)];
> +        /* HYP is an exception in that it is copied from r14 */
> +        if (mode == ARM_CPU_MODE_HYP) {
> +            env->xregs[14] = env->regs[14];
> +        } else {
> +            env->xregs[14] = env->banked_r14[bank_number(ARM_CPU_MODE_USR)];
> +        }
>      }

This part of the patch is a mess to read. I suspect this is a
combination of (a) your git not being configured to use a better
diff algorithm than the default (try "algorithm = histogram"
in the [diff] section of your .gitconfig), and it doing an effective
revert of 593cfa2b637b92d37 by accident.

It's also an absolutely enormous patch, even for a "just
moving code" patch, which makes it hard to review even
with diff --color-moved. Maybe it would be better in two
pieces ("helper routines for various insns" and "exception
handling functions" seems like a workable split).

thanks
-- PMM