[igt-dev] [PATCH i-g-t v2 3/3] tests/device_reset: Add cold reset IGT test
Iddamsetty, Aravind
aravind.iddamsetty at intel.com
Thu Nov 17 09:43:51 UTC 2022
On 16-11-2022 13:53, Anshuman Gupta wrote:
> Add cold-reset IGT subtest, IGT subtest will use
> /sys/bus/pci/slots/$SUN/power sysfs in order to
> initiate a cold reset sequence. $SUN value will be
> retrieved from PCIe device ACPI firmware node.
>
> v2:
> - %s/igt_require(fd > 0)/igt_assert(errno >0)
>
> Signed-off-by: Anshuman Gupta <anshuman.gupta at intel.com>
> Reviewed-by: Badal Nilawar <badal.nilawar at intel.com>
> ---
> lib/igt_pm.c | 40 ++++++++++++++
> lib/igt_pm.h | 1 +
> tests/device_reset.c | 124 +++++++++++++++++++++++++++++++++++++++----
> 3 files changed, 155 insertions(+), 10 deletions(-)
>
> diff --git a/lib/igt_pm.c b/lib/igt_pm.c
> index 4f0cfbdd10..9b7ed4be64 100644
> --- a/lib/igt_pm.c
> +++ b/lib/igt_pm.c
> @@ -877,6 +877,46 @@ static int igt_pm_open_pci_firmware_node(struct pci_device *pci_dev)
> return dir;
> }
>
> +/**
> + * igt_pm_get_pcie_acpihp_slot:
> + * @pci_dev: pci bridge device.
> + * Get pci bridge acpi hotplug slot number, if bridge's ACPI firmware_node
> + * handle supports _SUN method.
> + *
> + * Returns:
> + * PCIe bridge Slot number.
> + * Returns -ENOENT number in case firmware_node/sun is not supported by the
> + * bridge.
> + */
> +int igt_pm_get_pcie_acpihp_slot(struct pci_device *pci_dev)
> +{
> + int firmware_node_fd, fd, n_read;
> + char sun[8];
> +
> + firmware_node_fd = igt_pm_open_pci_firmware_node(pci_dev);
> +
> + if (firmware_node_fd < 0 && errno == ENOENT)
> + return -ENOENT;
> +
> + igt_require(firmware_node_fd > 0);
> +
> + fd = openat(firmware_node_fd, "sun", O_RDONLY);
> + if (fd < 0 && errno == ENOENT) {
> + close(firmware_node_fd);
> + return -ENOENT;
> + }
> +
> + igt_assert(errno > 0);
> +
> + n_read = read(fd, sun, sizeof(sun));
> + igt_assert(n_read > 0);
> +
> + close(firmware_node_fd);
> + close(fd);
> +
> + return strtol(sun, NULL, 10);
> +}
> +
> /**
> * igt_pm_acpi_d3cold_supported:
> * @pci_dev: root port pci_dev.
> diff --git a/lib/igt_pm.h b/lib/igt_pm.h
> index e81fceb897..f65b960c38 100644
> --- a/lib/igt_pm.h
> +++ b/lib/igt_pm.h
> @@ -73,6 +73,7 @@ bool igt_wait_for_pm_status(enum igt_runtime_pm_status status);
> bool igt_pm_dmc_loaded(int debugfs);
> bool igt_pm_pc8_plus_residencies_enabled(int msr_fd);
> bool i915_output_is_lpsp_capable(int drm_fd, igt_output_t *output);
> +int igt_pm_get_pcie_acpihp_slot(struct pci_device *pci_dev);
> bool igt_pm_acpi_d3cold_supported(struct pci_device *pci_dev);
> enum igt_acpi_d_state
> igt_pm_get_acpi_real_d_state(struct pci_device *pci_dev);
> diff --git a/tests/device_reset.c b/tests/device_reset.c
> index 0c477a02c0..427ea0b09f 100644
> --- a/tests/device_reset.c
> +++ b/tests/device_reset.c
> @@ -9,6 +9,7 @@
>
> #include "i915/gem.h"
> #include "igt.h"
> +#include "igt_device.h"
> #include "igt_device_scan.h"
> #include "igt_sysfs.h"
> #include "igt_kmod.h"
> @@ -33,6 +34,7 @@ struct device_fds {
> int dev;
> int dev_dir;
> int drv_dir;
> + int slot_dir; /* pci hotplug slots fd */
> } fds;
> char dev_bus_addr[DEV_BUS_ADDR_LEN];
> bool snd_unload;
> @@ -62,6 +64,45 @@ static int open_driver_sysfs_dir(int fd)
> return __open_sysfs_dir(fd, "device/driver");
> }
>
> +static int open_slot_sysfs_dir(int fd)
> +{
> + struct pci_device *pci_dev = NULL;
> + int slot_fd = -1, slot;
> + char slot_fd_path[PATH_MAX];
> +
> + pci_dev = igt_device_get_pci_device(fd);
> + igt_require(pci_dev);
> +
> + while ((pci_dev = pci_device_get_parent_bridge(pci_dev))) {
> + slot = igt_pm_get_pcie_acpihp_slot(pci_dev);
> + if (slot == -ENOENT) {
> + igt_debug("Bridge PCI device %04x:%02x:%02x.%01x does not support acpihp slot\n",
> + pci_dev->domain, pci_dev->bus, pci_dev->dev, pci_dev->func);
> + continue;
> + }
> +
> + /*
> + * Upon getting the valid acpihp slot number break the loop.
> + * It is the desired acpihp slot for gfx card.
> + */
> + if (slot > 0) {
> + igt_debug("Bridge PCI device %04x:%02x:%02x.%01x associated acpihp slot %d\n",
> + pci_dev->domain, pci_dev->bus, pci_dev->dev, pci_dev->func, slot);
> + break;
> + }
> + }
> +
> + if (!pci_dev)
> + return slot_fd;
> +
> + snprintf(slot_fd_path, PATH_MAX, "/sys/bus/pci/slots/%d", slot);
> + slot_fd = open(slot_fd_path, O_RDONLY);
> + if (slot_fd < 0)
> + return -errno;
> +
> + return slot_fd;
> +}
> +
> /**
> * device_sysfs_path:
> * @fd: opened device file descriptor
> @@ -124,6 +165,8 @@ static void init_device_fds(struct device_fds *dev)
>
> dev->fds.drv_dir = open_driver_sysfs_dir(dev->fds.dev);
> igt_assert_fd(dev->fds.drv_dir);
> +
> + dev->fds.slot_dir = open_slot_sysfs_dir(dev->fds.dev);
> }
>
> static int close_if_opened(int *fd)
> @@ -142,6 +185,7 @@ static void cleanup_device_fds(struct device_fds *dev)
> igt_ignore_warn(close_if_opened(&dev->fds.dev));
> igt_ignore_warn(close_if_opened(&dev->fds.dev_dir));
> igt_ignore_warn(close_if_opened(&dev->fds.drv_dir));
> + igt_ignore_warn(close_if_opened(&dev->fds.slot_dir));
> }
>
> /**
> @@ -179,6 +223,34 @@ static bool is_sysfs_reset_supported(int fd)
> return true;
> }
>
> +/**
> + * is_sysfs_cold_reset_supported:
> + * @fd: opened device file descriptor
> + *
> + * Check if device supports cold reset based on sysfs file presence.
> + *
> + * Returns:
> + * True if device supports reset, false otherwise.
> + */
I do not think just with the presence of power sysfs entry we can
confirm that slot supports cold reset or in fact it can be powered
down/up. we should even check for slot capabilites register for Power
Controller Present bit.
Thanks,
Aravind.
> +static bool is_sysfs_cold_reset_supported(int slot_fd)
> +{
> + struct stat st;
> + int rc;
> + int cold_reset_fd = -1;
> +
> + cold_reset_fd = openat(slot_fd, "power", O_WRONLY);
> +
> + if (cold_reset_fd < 0)
> + return false;
> +
> + rc = fstat(cold_reset_fd, &st);
> + close(cold_reset_fd);
> +
> + if (rc || !S_ISREG(st.st_mode))
> + return false;
> +
> + return true;
> +}
> /* Unbind the driver from the device */
> static void driver_unbind(struct device_fds *dev)
> {
> @@ -231,8 +303,12 @@ static void initiate_device_reset(struct device_fds *dev, enum reset type)
> {
> igt_debug("reset device\n");
>
> - if (type == FLR_RESET)
> + if (type == FLR_RESET) {
> igt_assert(igt_sysfs_set(dev->fds.dev_dir, "reset", "1"));
> + } else if (type == COLD_RESET) {
> + igt_assert(igt_sysfs_set(dev->fds.slot_dir, "power", "0"));
> + igt_assert(igt_sysfs_set(dev->fds.slot_dir, "power", "1"));
> + }
>
> }
>
> @@ -311,17 +387,45 @@ igt_main
> igt_skip_on(!is_sysfs_reset_supported(dev.fds.dev));
> }
>
> - igt_describe("Unbinds driver from device, initiates reset"
> - " then rebinds driver to device");
> - igt_subtest("unbind-reset-rebind") {
> - unbind_reset_rebind(&dev, FLR_RESET);
> - healthcheck(&dev);
> + igt_subtest_group {
> + igt_describe("Unbinds driver from device, initiates reset"
> + " then rebinds driver to device");
> + igt_subtest("unbind-reset-rebind") {
> + unbind_reset_rebind(&dev, FLR_RESET);
> + healthcheck(&dev);
> + }
> +
> + igt_describe("Resets device with bound driver");
> + igt_subtest("reset-bound") {
> + initiate_device_reset(&dev, FLR_RESET);
> + /*
> + * Cold reset will initiate card boot sequence again,
> + * therefore let healthcheck() re-epen the dev fd.
> + */
> + dev.fds.dev = -1;
> + healthcheck(&dev);
> + }
> }
>
> - igt_describe("Resets device with bound driver");
> - igt_subtest("reset-bound") {
> - initiate_device_reset(&dev, FLR_RESET);
> - healthcheck(&dev);
> + igt_subtest_group {
> + igt_fixture {
> + igt_skip_on_f(dev.fds.slot_dir < 0, "Gfx Card does not support any "
> + "pcie slot for cold reset\n");
> + igt_skip_on(!is_sysfs_cold_reset_supported(dev.fds.slot_dir));
> + }
> +
> + igt_describe("Unbinds driver from device, initiates cold reset"
> + " then rebinds driver to device");
> + igt_subtest("unbind-cold-reset-rebind") {
> + unbind_reset_rebind(&dev, COLD_RESET);
> + healthcheck(&dev);
> + }
> +
> + igt_describe("Cold Resets device with bound driver");
> + igt_subtest("cold-reset-bound") {
> + initiate_device_reset(&dev, COLD_RESET);
> + healthcheck(&dev);
> + }
> }
>
> igt_fixture {
More information about the igt-dev
mailing list