[PATCH v2] arm64: dts: renesas: r8a78000: Fix GIC-720AE View 1 Redistributor description

[PATCH v2] arm64: dts: renesas: r8a78000: Fix GIC-720AE View 1 Redistributor description

[Marek Vasut]

The Renesas R-Car X5H (R8A78000) SoC contains Arm CoreLink GIC-720AE
Generic Interrupt Controller with Multi View capability. Firmware has
access to configuration View 0, Linux kernel has access to View 1.

The Arm CoreLink GIC-720AE Generic Interrupt Controller Technical
Reference Manual, currently latest r2p1 [1], chapter "5. Programmers
model for GIC-720AE", subchapter "5.4 Redistributor registers
for control and physical LPIs summary", part "5.4.3 GICR_TYPER,
Redistributor Type Register", "Table 5-50: GICR_TYPER bit descriptions"
on page 200, clarifies register "GICR_TYPER" bit 4 "Last" behavior
in Multi View setup as follows:

"
Last
Last Redistributor:

0 ... This Redistributor is not the last Redistributor on the chip.
1 ... This Redistributor is the last Redistributor on the chip.
      When GICD_CFGID.VIEW == 1, for views 1, 2, or 3 this bit
      always returns 1.
"

On this SoC, GICD_CFGID.VIEW is 1 and the Linux kernel has access to
View 1, therefore Linux kernel GICv3 driver will interpret register
"GICR_TYPER" bit 4 "Last" = 1 in the first Redistributor in continuous
Redistributor page as that first Redistributor being the one and only
Redistributor and will stop processing the continuous Redistributor
page further. This will prevent the other Redistributors from being
recognized by the system and used for other PEs.

Because the hardware indicates that the continuous Redistributor page
is not continuous for View 1, 2, or 3, describe every Redistributor
separately in the DT. This makes all Redistributors for all cores
accessible in Linux.

[1] https://documentation-service.arm.com/static/69ef3c1cd35efd294e335c43
    Arm® CoreLink™ GIC-720AE Generic Interrupt Controller
    Revision: r2p1 / Issue 12 / 102666_0201_12_en

Fixes: 63500d12cf76 ("arm64: dts: renesas: Add R8A78000 SoC support")
Acked-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
---
Cc: Conor Dooley <conor+dt@kernel.org>
Cc: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: Krzysztof Kozlowski <krzk+dt@kernel.org>
Cc: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Cc: Magnus Damm <magnus.damm@gmail.com>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Rob Herring <robh@kernel.org>
Cc: devicetree@vger.kernel.org
Cc: linux-renesas-soc@vger.kernel.org
---
V2: - Update GIC-700AE documentation link to generated PDF
    - Update commit message to match the PDF
    - Add AB from Marc
---
 arch/arm64/boot/dts/renesas/r8a78000.dtsi | 36 +++++++++++++++++++++--
 1 file changed, 34 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/boot/dts/renesas/r8a78000.dtsi b/arch/arm64/boot/dts/renesas/r8a78000.dtsi
index 3ec1b53d27828..73be51787265c 100644
--- a/arch/arm64/boot/dts/renesas/r8a78000.dtsi
+++ b/arch/arm64/boot/dts/renesas/r8a78000.dtsi
@@ -689,8 +689,40 @@ gic: interrupt-controller@39000000 {
 			#interrupt-cells = <3>;
 			#address-cells = <0>;
 			interrupt-controller;
-			reg = <0 0x39000000 0 0x10000>,
-			      <0 0x39080000 0 0x800000>;
+			#redistributor-regions = <32>;
+			reg = <0x0 0x39000000 0x0 0x10000>,
+			      <0x0 0x39080000 0x0 0x40000>,
+			      <0x0 0x390c0000 0x0 0x40000>,
+			      <0x0 0x39100000 0x0 0x40000>,
+			      <0x0 0x39140000 0x0 0x40000>,
+			      <0x0 0x39180000 0x0 0x40000>,
+			      <0x0 0x391c0000 0x0 0x40000>,
+			      <0x0 0x39200000 0x0 0x40000>,
+			      <0x0 0x39240000 0x0 0x40000>,
+			      <0x0 0x39280000 0x0 0x40000>,
+			      <0x0 0x392c0000 0x0 0x40000>,
+			      <0x0 0x39300000 0x0 0x40000>,
+			      <0x0 0x39340000 0x0 0x40000>,
+			      <0x0 0x39380000 0x0 0x40000>,
+			      <0x0 0x393c0000 0x0 0x40000>,
+			      <0x0 0x39400000 0x0 0x40000>,
+			      <0x0 0x39440000 0x0 0x40000>,
+			      <0x0 0x39480000 0x0 0x40000>,
+			      <0x0 0x394c0000 0x0 0x40000>,
+			      <0x0 0x39500000 0x0 0x40000>,
+			      <0x0 0x39540000 0x0 0x40000>,
+			      <0x0 0x39580000 0x0 0x40000>,
+			      <0x0 0x395c0000 0x0 0x40000>,
+			      <0x0 0x39600000 0x0 0x40000>,
+			      <0x0 0x39640000 0x0 0x40000>,
+			      <0x0 0x39680000 0x0 0x40000>,
+			      <0x0 0x396c0000 0x0 0x40000>,
+			      <0x0 0x39700000 0x0 0x40000>,
+			      <0x0 0x39740000 0x0 0x40000>,
+			      <0x0 0x39780000 0x0 0x40000>,
+			      <0x0 0x397c0000 0x0 0x40000>,
+			      <0x0 0x39800000 0x0 0x40000>,
+			      <0x0 0x39840000 0x0 0x40000>;
 			interrupts = <GIC_PPI 9 IRQ_TYPE_LEVEL_HIGH>;
 		};
 
-- 
2.53.0

Re: [PATCH v2] arm64: dts: renesas: r8a78000: Fix GIC-720AE View 1 Redistributor description

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] DT fragmentation creates a booby trap leading to a fatal Data Abort if booted in View 0.
- [High] Splitting regions makes the GIC probe extremely brittle to sparse CPU topologies.
--

commit ff8b7280cc2b6f0b2be6c7d236cd18a4ed49d89b
Author: Marek Vasut <marek.vasut+renesas@mailbox.org>

arm64: dts: renesas: r8a78000: Fix GIC-720AE View 1 Redistributor description

This commit splits the contiguous 8MB redistributor region into 32 separate
256KB regions. This attempts to work around the GIC-720AE View 1 behavior
where the first redistributor returns GICR_TYPER.Last == 1.

> diff --git a/arch/arm64/boot/dts/renesas/r8a78000.dtsi b/arch/arm64/boot/dts/renesas/r8a78000.dtsi
> index 4c97298fa7634..fd268604c0ec5 100644
> --- a/arch/arm64/boot/dts/renesas/r8a78000.dtsi
> +++ b/arch/arm64/boot/dts/renesas/r8a78000.dtsi
> @@ -689,8 +689,40 @@ gic: interrupt-controller@39000000 {
>  			#interrupt-cells = <3>;
>  			#address-cells = <0>;
>  			interrupt-controller;
> -			reg = <0 0x39000000 0 0x10000>,
> -			      <0 0x39080000 0 0x800000>;
> +			#redistributor-regions = <32>;
> +			reg = <0x0 0x39000000 0x0 0x10000>,
> +			      <0x0 0x39080000 0x0 0x40000>,
> +			      <0x0 0x390c0000 0x0 0x40000>,

Could this cause a Data Abort if the system is booted in View 0?

Because the Device Tree binding doesn't currently support setting
single_redist = true for these regions, each 256KB region is mapped exactly
to that size via ioremap.

If booted in View 0, the hardware operates normally and intermediate
redistributors return GICR_TYPER.Last == 0.

The GIC driver's gic_iterate_rdists() inner loop relies entirely on the
GICR_TYPER.Last == 1 bit to break. Without it, the driver will advance ptr
and attempt to read gic_read_typer(ptr + GICR_TYPER) exactly at or beyond the
256KB ioremap boundary, triggering a Data Abort and kernel panic during early
boot.

Would it be safer to keep the DT as a single continuous region and add a
hardware-specific quirk to the GIC driver to bypass the LAST bit?

Additionally, does splitting these regions make the GIC probe brittle on
sparse CPU topologies?

If a redistributor is missing, such as a fused-off core, and returns an
invalid PIDR2, the driver will issue a warning and execute a break.

This break exits the entire outer for loop in gic_iterate_rdists(),
immediately aborting the discovery of all subsequent redistributor regions.

Any present CPUs located after a fused-off core would then fail to be
probed and will fail to boot.

[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260514125328.20954-1-marek.vasut+renesas@mailbox.org?part=1