From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:37946) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1azp0g-00040k-Ni for qemu-devel@nongnu.org; Mon, 09 May 2016 13:30:23 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1azp0e-0000Hw-P7 for qemu-devel@nongnu.org; Mon, 09 May 2016 13:30:21 -0400 From: Peter Maydell Date: Mon, 9 May 2016 18:29:47 +0100 Message-Id: <1462814989-24360-22-git-send-email-peter.maydell@linaro.org> In-Reply-To: <1462814989-24360-1-git-send-email-peter.maydell@linaro.org> References: <1462814989-24360-1-git-send-email-peter.maydell@linaro.org> Subject: [Qemu-devel] [PATCH 21/23] hw/intc/arm_gicv3: Work around Linux assuming interrupts are group 1 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-arm@nongnu.org, qemu-devel@nongnu.org Cc: patches@linaro.org, Pavel Fedin , Shlomo Pongratz , Shlomo Pongratz , Christoffer Dall , Shannon Zhao The Linux kernel's GICv3 driver assumes that all interrupts are in group 1. This is correct if the system supports the Security extensions, because in that case the kernel cannot configure the interrupts and it must have been done already by firmware. However if the system does not support the Security extensions then the kernel is perfectly capable of configuring them into group 1 itself if it wants them there; it just doesn't. Work around this by having the GICv3 emulation put all the interrupts into group 1 if we're directly booting a Linux kernel, whether the Security extensions are supported or not. Signed-off-by: Peter Maydell --- hw/intc/arm_gicv3_common.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/hw/intc/arm_gicv3_common.c b/hw/intc/arm_gicv3_common.c index 901ec60..73d3c6d 100644 --- a/hw/intc/arm_gicv3_common.c +++ b/hw/intc/arm_gicv3_common.c @@ -288,6 +288,13 @@ static void arm_gic_common_linux_init(ARMLinuxBootIf *obj, * equivalent). */ s->irq_reset_nonsecure = true; + } else { + /* This is purely a workaround for broken Linux kernel behaviour + * on non-TrustZone systems. It assumes that interrupts have been + * set to group 1 even though it could do that itself for a non-secure + * GIC. + */ + s->irq_reset_nonsecure = true; } } -- 1.9.1