[v1] ARM AArch64 Support for BSD

[PATCH 23/23] Add get_ucontext_sigreturn function

Posted by Ajeet Singh 5 months, 1 week ago

From: Stacey Son <sson@FreeBSD.org>

Function checks the processor state to ensure that the current
execution mode is EL0 and no flags indicating interrupts or
exceptions are set

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Ajeet Singh <itachis@FreeBSD.org>
---
 bsd-user/aarch64/signal.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/bsd-user/aarch64/signal.c b/bsd-user/aarch64/signal.c
index 13faac8ce6..ad81531ec5 100644
--- a/bsd-user/aarch64/signal.c
+++ b/bsd-user/aarch64/signal.c
@@ -117,3 +117,21 @@ abi_long set_mcontext(CPUARMState *regs, target_mcontext_t *mcp, int srflag)
 
     return err;
 }
+
+/* Compare to sys_sigreturn() in  arm64/arm64/machdep.c */
+abi_long get_ucontext_sigreturn(CPUARMState *regs, abi_ulong target_sf,
+                                abi_ulong *target_uc)
+{
+    uint32_t pstate = pstate_read(regs);
+
+    *target_uc = 0;
+
+    if ((pstate & PSTATE_M) != PSTATE_MODE_EL0t  ||
+        (pstate & (PSTATE_F | PSTATE_I | PSTATE_A | PSTATE_D)) != 0) {
+        return -TARGET_EINVAL;
+    }
+
+    *target_uc = target_sf;
+
+    return 0;
+}
-- 
2.34.1

Re: [PATCH 23/23] Add get_ucontext_sigreturn function

Posted by Richard Henderson 5 months, 1 week ago

On 6/17/24 11:58, Ajeet Singh wrote:
> From: Stacey Son <sson@FreeBSD.org>
> 
> Function checks the processor state to ensure that the current
> execution mode is EL0 and no flags indicating interrupts or
> exceptions are set
> 
> Signed-off-by: Stacey Son <sson@FreeBSD.org>
> Signed-off-by: Ajeet Singh <itachis@FreeBSD.org>
> ---
>   bsd-user/aarch64/signal.c | 18 ++++++++++++++++++
>   1 file changed, 18 insertions(+)
> 
> diff --git a/bsd-user/aarch64/signal.c b/bsd-user/aarch64/signal.c
> index 13faac8ce6..ad81531ec5 100644
> --- a/bsd-user/aarch64/signal.c
> +++ b/bsd-user/aarch64/signal.c
> @@ -117,3 +117,21 @@ abi_long set_mcontext(CPUARMState *regs, target_mcontext_t *mcp, int srflag)
>   
>       return err;
>   }
> +
> +/* Compare to sys_sigreturn() in  arm64/arm64/machdep.c */
> +abi_long get_ucontext_sigreturn(CPUARMState *regs, abi_ulong target_sf,
> +                                abi_ulong *target_uc)
> +{
> +    uint32_t pstate = pstate_read(regs);
> +
> +    *target_uc = 0;
> +
> +    if ((pstate & PSTATE_M) != PSTATE_MODE_EL0t  ||
> +        (pstate & (PSTATE_F | PSTATE_I | PSTATE_A | PSTATE_D)) != 0) {
> +        return -TARGET_EINVAL;
> +    }
> +
> +    *target_uc = target_sf;

Why delay this store?  I don't see why you're assigning 0 above.


r~

Re: [PATCH 23/23] Add get_ucontext_sigreturn function

Posted by Warner Losh 5 months ago

On Tue, Jun 18, 2024 at 4:56 PM Richard Henderson <
richard.henderson@linaro.org> wrote:

> On 6/17/24 11:58, Ajeet Singh wrote:
> > From: Stacey Son <sson@FreeBSD.org>
> >
> > Function checks the processor state to ensure that the current
> > execution mode is EL0 and no flags indicating interrupts or
> > exceptions are set
> >
> > Signed-off-by: Stacey Son <sson@FreeBSD.org>
> > Signed-off-by: Ajeet Singh <itachis@FreeBSD.org>
> > ---
> >   bsd-user/aarch64/signal.c | 18 ++++++++++++++++++
> >   1 file changed, 18 insertions(+)
> >
> > diff --git a/bsd-user/aarch64/signal.c b/bsd-user/aarch64/signal.c
> > index 13faac8ce6..ad81531ec5 100644
> > --- a/bsd-user/aarch64/signal.c
> > +++ b/bsd-user/aarch64/signal.c
> > @@ -117,3 +117,21 @@ abi_long set_mcontext(CPUARMState *regs,
> target_mcontext_t *mcp, int srflag)
> >
> >       return err;
> >   }
> > +
> > +/* Compare to sys_sigreturn() in  arm64/arm64/machdep.c */
>

This is now in exec_machdep.c and the most relevant bits are set_mcontext()
and sys_sigreturn().


> > +abi_long get_ucontext_sigreturn(CPUARMState *regs, abi_ulong target_sf,
> > +                                abi_ulong *target_uc)
> > +{
> > +    uint32_t pstate = pstate_read(regs);
> > +
> > +    *target_uc = 0;
> > +
> > +    if ((pstate & PSTATE_M) != PSTATE_MODE_EL0t  ||
> > +        (pstate & (PSTATE_F | PSTATE_I | PSTATE_A | PSTATE_D)) != 0) {
> > +        return -TARGET_EINVAL;
> > +    }
> > +
> > +    *target_uc = target_sf;
>
> Why delay this store?  I don't see why you're assigning 0 above.
>

I'm not sure I understand this either....  We don't store anything when
there's an error in the pstate, at least in the kernel code.

Warner