[igt-dev] [PATCH i-g-t v2 2/2] tests/i915/i915_suspend: Disable d3cold_allowed for basic-s2idle-without-i915

[Riana Tauro]

When module is unloaded and s2idle is triggered,
PCI core leaves the endpoint in D0 and the bridge in D3 state causing
PCIE spec violation. The config space is read as 0xFF

Keep the bridge in D0 before module unload to prevent
this issue

v2: Addressed cosmetic review comments (Anshuman)

Signed-off-by: Riana Tauro <riana.tauro at intel.com>
---
 tests/i915/i915_suspend.c | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/tests/i915/i915_suspend.c b/tests/i915/i915_suspend.c
index 94935595..815f1c8a 100644
--- a/tests/i915/i915_suspend.c
+++ b/tests/i915/i915_suspend.c
@@ -43,6 +43,7 @@
 #include "igt.h"
 #include "igt_kmod.h"
 #include "igt_device.h"
+#include "igt_device_scan.h"
 
 #define OBJECT_SIZE (16*1024*1024)
 
@@ -206,13 +207,38 @@ test_forcewake(int fd, bool hibernate)
 static void
 test_suspend_without_i915(int state)
 {
+	struct igt_device_card card;
+	char d3cold_allowed[2];
+	int fd;
+
+	fd = __drm_open_driver(DRIVER_INTEL);
+	igt_devices_scan(false);
+
+	/*
+	 * When module is unloaded and s2idle is triggered, PCI core leaves the endpoint
+	 * in D0 and the bridge in D3 state causing PCIE spec violation and config space
+	 * is read as 0xFF. Keep the bridge in D0 before module unload to prevent
+	 * this issue
+	 */
+	if (state == SUSPEND_STATE_FREEZE &&
+	    igt_device_find_first_i915_discrete_card(&card)) {
+		igt_pm_get_d3cold_allowed(&card, d3cold_allowed);
+		igt_pm_set_d3cold_allowed(&card, "0\n");
+	}
+	close(fd);
+
 	igt_kmsg(KMSG_INFO "Unloading i915\n");
 	igt_assert_eq(igt_i915_driver_unload(),0);
 
 	igt_system_suspend_autoresume(state, SUSPEND_TEST_NONE);
 
+	if (state == SUSPEND_STATE_FREEZE && strlen(card.card))
+		igt_pm_set_d3cold_allowed(&card, d3cold_allowed);
+
 	igt_kmsg(KMSG_INFO "Re-loading i915 \n");
 	igt_assert_eq(igt_i915_driver_load(NULL), 0);
+
+	igt_devices_free();
 }
 
 int fd;
-- 
2.25.1