Re: [kvm-unit-tests PATCH v2 09/16] arm/arm64: ITS: Enable/Disable LPIs at re-distributor level

From: Andrew Jones <drjones@redhat.com>
To: Eric Auger <eric.auger@redhat.com>
Cc: thuth@redhat.com, kvm@vger.kernel.org, maz@kernel.org,
	qemu-devel@nongnu.org, qemu-arm@nongnu.org,
	andre.przywara@arm.com, kvmarm@lists.cs.columbia.edu,
	eric.auger.pro@gmail.com
Subject: Re: [kvm-unit-tests PATCH v2 09/16] arm/arm64: ITS: Enable/Disable LPIs at re-distributor level
Date: Mon, 13 Jan 2020 18:44:57 +0100	[thread overview]
Message-ID: <20200113174457.gg35yyeaftbadqki@kamzik.brq.redhat.com> (raw)
In-Reply-To: <20200110145412.14937-10-eric.auger@redhat.com>

On Fri, Jan 10, 2020 at 03:54:05PM +0100, Eric Auger wrote:
> This helper function enables or disables the signaling of LPIs
> at redistributor level.
> 
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> ---
>  lib/arm/asm/gic-v3-its.h |  1 +
>  lib/arm/gic-v3-its.c     | 18 ++++++++++++++++++
>  2 files changed, 19 insertions(+)
> 
> diff --git a/lib/arm/asm/gic-v3-its.h b/lib/arm/asm/gic-v3-its.h
> index 93814f7..d2db292 100644
> --- a/lib/arm/asm/gic-v3-its.h
> +++ b/lib/arm/asm/gic-v3-its.h
> @@ -99,6 +99,7 @@ extern struct its_baser *its_lookup_baser(int type);
>  extern void set_lpi_config(int n, u8 val);
>  extern u8 get_lpi_config(int n);
>  extern void set_pending_table_bit(int rdist, int n, bool set);
> +extern void gicv3_rdist_ctrl_lpi(u32 redist, bool set);
>  
>  #endif /* !__ASSEMBLY__ */
>  #endif /* _ASMARM_GIC_V3_ITS_H_ */
> diff --git a/lib/arm/gic-v3-its.c b/lib/arm/gic-v3-its.c
> index 3037c84..c7c6f80 100644
> --- a/lib/arm/gic-v3-its.c
> +++ b/lib/arm/gic-v3-its.c
> @@ -199,3 +199,21 @@ void init_cmd_queue(void)
>  	its_data.cmd_write = its_data.cmd_base;
>  	writeq(0, its_data.base + GITS_CWRITER);
>  }
> +
> +void gicv3_rdist_ctrl_lpi(u32 redist, bool set)
> +{
> +	void *ptr;
> +	u64 val;
> +
> +	if (redist >= nr_cpus)
> +		report_abort("%s redist=%d >= cpu_count=%d\n",
> +			     __func__, redist, nr_cpus);

I'd use {} here because of the multiline call. But, we don't
use the report API in common code. Well, apparently s390 has
report calls in lib/s390x/interrupt.c, but I don't really
agree with that. IMO, common code failures should always
be unexpected and just assert/assert_msg.

> +
> +	ptr = gicv3_data.redist_base[redist];
> +	val = readl(ptr + GICR_CTLR);
> +	if (set)
> +		val |= GICR_CTLR_ENABLE_LPIS;
> +	else
> +		val &= ~GICR_CTLR_ENABLE_LPIS;
> +	writel(val,  ptr + GICR_CTLR);
> +}
> -- 
> 2.20.1
> 

Also, you can squash this patch into whatever is going to make use
of this new helper.

Thanks,
drew

_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm