[SLOF] [PATCH 2/2] board-qemu: add private hcall to inform host on "phandle" update

[Thomas Huth]

On 10.07.2017 14:17, Greg Kurz wrote:
> The "interrupt-map" property in each PHB node references the "phandle"
> property of the "interrupt-controller" node. This is used by the guest
> OS to setup IRQs for any PCI device plugged into the PHB. QEMU sets this
> property to an arbitrary value in the flattened DT passed to SLOF.
> 
> Since commit 82954d4c1088, SLOF has some generic code to convert all
> references to any "phandle" property to a SLOF specific value.
> 
> This is is perfectly okay for coldplug devices, since the guest OS only
> sees the converted value in "interrupt-map". It is a problem though for
> hotplug devices. Since they don't go through SLOF, the guest OS receives
> the arbitrary value set by QEMU and fails to setup IRQs.
> 
> In order to support PHB hotplug, this patch introduces a new private
> hcall, which allows SLOF to tell QEMU that a "phandle" was converted
> from an old value to a new value.
> 
> Suggested-by: Thomas Huth <thuth at redhat.com>
> Signed-off-by: Greg Kurz <groug at kaod.org>
> ---
>  board-qemu/slof/fdt.fs    |   14 +++++++++++---
>  lib/libhvcall/hvcall.code |    6 ++++++
>  lib/libhvcall/hvcall.in   |    1 +
>  lib/libhvcall/libhvcall.h |    1 +
>  4 files changed, 19 insertions(+), 3 deletions(-)
> 
> diff --git a/board-qemu/slof/fdt.fs b/board-qemu/slof/fdt.fs
> index 8d4635f30495..8ce0a59879ac 100644
> --- a/board-qemu/slof/fdt.fs
> +++ b/board-qemu/slof/fdt.fs
> @@ -316,14 +316,22 @@ fdt-claim-reserve
>     s" phandle" r@ get-property 0= IF
>        decode-int                       ( p-addr2 p-len2 val )
>        \ ." found phandle: " dup . cr
> -      r@ s" /" find-node               ( p-addr2 p-len2 val node root )  
> -      fdt-replace-all-phandles         ( p-addr2 p-len2 )
> -      2drop
> +      dup                              ( p-addr2 p-len2 val val )
> +      r@ s" /" find-node               ( p-addr2 p-len2 val val node root )
> +      fdt-replace-all-phandles         ( p-addr2 p-len2 val )
> +      -rot 2drop                       ( val )
>        (fdt-phandle-replaced) IF
>           r@ set-node
>           s" phandle" delete-property
>           s" linux,phandle" delete-property
> +         r@ swap hv-update-phandle dup 0 <> IF

Instead of "dup 0 <> IF" you can use "?dup IF" here and remove the
corresponding "ELSE drop" below.

> +	    \ Ignore hcall not implemented error, print error otherwise
> +	    dup -2 <> IF ." HV-UPDATE-PHANDLE error: " . cr ELSE drop THEN

Dito - use ?dup and remove the "ELSE drop" part.

> +         ELSE
> +	    drop
> +         THEN
>        ELSE
> +         drop
>           diagnostic-mode? IF
>              cr ." Warning: Did not replace phandle in " r@ node>path type cr
>           THEN
> diff --git a/lib/libhvcall/hvcall.code b/lib/libhvcall/hvcall.code
> index 0ff50f27e8a9..9f4d53c7237b 100644
> --- a/lib/libhvcall/hvcall.code
> +++ b/lib/libhvcall/hvcall.code
> @@ -129,3 +129,9 @@ PRIM(check_X2d_and_X2d_patch_X2d_sc1)
>  
>  	patch_broken_sc1((void*)start, (void*)end, (void*)patch_ins);
>  MIRP

Could you add a stack comment before that function, please? (like it is
done for hv-reg-crq for example already)

> +PRIM(hv_X2d_update_X2d_phandle)
> +	uint32_t old_phandle = TOS.u; POP;
> +	uint32_t new_phandle = TOS.u;

If you switch the order of the parameters, you could get rid of the
"swap" before calling this function :-)

> +	TOS.u = hv_generic(KVMPPC_H_UPDATE_PHANDLE, old_phandle, new_phandle);
> +MIRP
> diff --git a/lib/libhvcall/hvcall.in b/lib/libhvcall/hvcall.in
> index 4437b77f001d..ab7513af8977 100644
> --- a/lib/libhvcall/hvcall.in
> +++ b/lib/libhvcall/hvcall.in
> @@ -31,4 +31,5 @@ cod(RX!)
>  cod(hv-logical-memop)
>  cod(hv-cas)
>  cod(hv-rtas-update)
> +cod(hv-update-phandle)
>  cod(get-print-version)
> diff --git a/lib/libhvcall/libhvcall.h b/lib/libhvcall/libhvcall.h
> index b2ea3f6bf944..5776a2b772f7 100644
> --- a/lib/libhvcall/libhvcall.h
> +++ b/lib/libhvcall/libhvcall.h
> @@ -25,6 +25,7 @@
>  /* Client Architecture support */
>  #define KVMPPC_H_CAS            (KVMPPC_HCALL_BASE + 0x2)
>  #define KVMPPC_H_RTAS_UPDATE    (KVMPPC_HCALL_BASE + 0x3)
> +#define KVMPPC_H_UPDATE_PHANDLE (KVMPPC_HCALL_BASE + 0x4)
>  
>  #ifndef __ASSEMBLY__

Apart from the nits, the patch looks fine to me.

 Thomas