[PATCH i-g-t v2] Add single engine busyness stats in GPUTOP
Belgaumkar, Vinay
vinay.belgaumkar at intel.com
Fri Feb 21 04:17:31 UTC 2025
On 2/14/2025 8:32 AM, Soham Purkait wrote:
> Add single engine busyness support in GPUTOP.
> This uses the PMU interface to display the
> busyness of each engine instances.
>
> ENGINES BUSY
> Render/3D/0 | 96.5% ███████████████████████████████████████▍|
> Blitter/0 | 91.6% █████████████████████████████████████ |
> Video/0 | 56.2% ███████████████████████████ |
> VideoEnhance/0| 97.7% ████████████████████████████████████████|
> Compute/0 | 48.5% ███████████████████████▍ |
>
> v1 : fixed cosmetic issues
>
> v2 : fix for refactoring GPUTOP into a
> vendor-agnostic tool (Lucas)
>
> ---
> lib/igt_device_scan.c | 82 ++++++++
> lib/igt_device_scan.h | 5 +
> lib/igt_perf.c | 53 ++++++
> lib/igt_perf.h | 2 +
> tools/gputop/common_gputop.c | 51 +++++
> tools/gputop/common_gputop.h | 16 ++
> tools/{ => gputop}/gputop.c | 246 ++++++++++++++++++++----
> tools/gputop/meson.build | 6 +
> tools/gputop/xe_gputop.c | 359 +++++++++++++++++++++++++++++++++++
> tools/gputop/xe_gputop.h | 74 ++++++++
> tools/meson.build | 6 +-
> 11 files changed, 858 insertions(+), 42 deletions(-)
> create mode 100644 tools/gputop/common_gputop.c
> create mode 100644 tools/gputop/common_gputop.h
> rename tools/{ => gputop}/gputop.c (65%)
> create mode 100644 tools/gputop/meson.build
> create mode 100644 tools/gputop/xe_gputop.c
> create mode 100644 tools/gputop/xe_gputop.h
>
> diff --git a/lib/igt_device_scan.c b/lib/igt_device_scan.c
> index 711bedc5c..c71db0094 100644
> --- a/lib/igt_device_scan.c
> +++ b/lib/igt_device_scan.c
> @@ -773,6 +773,9 @@ __copy_dev_to_card(struct igt_device *dev, struct igt_device_card *card)
> if (dev->drm_render != NULL)
> safe_strncpy(card->render, dev->drm_render,
> sizeof(card->render));
> + if (dev->driver != NULL)
> + safe_strncpy(card->driver, dev->driver,
> + sizeof(card->driver));
>
> if (dev->pci_slot_name != NULL)
> safe_strncpy(card->pci_slot_name, dev->pci_slot_name,
> @@ -820,6 +823,61 @@ static bool __find_first_intel_card_by_driver_name(struct igt_device_card *card,
> return false;
> }
>
> +/*
> + * Iterate over all igt_devices array and find all discrete/integrated card.
> + * @card: double pointer to igt_device_card structure, containing
> + * an array of igt_device_card structure upon successful return.
> + */
> +static int __find_all_intel_card_by_driver_name(struct igt_device_card **card,
> + bool want_discrete, const char *drv_name)
> +{
> + int count = 0;
> + struct igt_device *dev;
> + int is_integrated;
> + struct igt_device_card *tmp;
> + struct igt_device_card *crd =
> + (struct igt_device_card *)calloc(1, sizeof(struct igt_device_card));
> +
> + igt_assert(drv_name);
> + memset(card, 0, sizeof(*card));
> +
> + igt_list_for_each_entry(dev, &igt_devs.all, link) {
> + if (!is_pci_subsystem(dev) || strcmp(dev->driver, drv_name))
> + continue;
> +
> + is_integrated = !strncmp(dev->pci_slot_name, INTEGRATED_I915_GPU_PCI_ID,
> + PCI_SLOT_NAME_SIZE);
> +
> + if (want_discrete && !is_integrated) {
> + __copy_dev_to_card(dev, (crd + count));
> + count++;
> + tmp = realloc(crd, sizeof(struct igt_device_card) * (1 + count));
> + if (!tmp) {
> + free(crd);
> + return -1;
> + }
> + crd = tmp;
> +
> + } else if (!want_discrete && is_integrated) {
> + __copy_dev_to_card(dev, (crd + count));
> + count++;
> + tmp = realloc(crd, sizeof(struct igt_device_card) * (1 + count));
> + if (!tmp) {
> + free(crd);
> + return -1;
> + }
> + crd = tmp;
> + }
> + }
> + if (count == 0) {
> + free(crd);
> + return 0;
> + }
> +
> + *card = crd;
> + return count;
> +}
> +
> bool igt_device_find_first_i915_discrete_card(struct igt_device_card *card)
> {
> igt_assert(card);
> @@ -866,6 +924,30 @@ bool igt_device_find_xe_integrated_card(struct igt_device_card *card)
> return __find_first_intel_card_by_driver_name(card, false, "xe");
> }
>
> +int igt_device_find_all_xe_integrated_card(struct igt_device_card **card)
> +{
> + igt_assert(card);
> + return __find_all_intel_card_by_driver_name(card, false, "xe");
> +}
> +
> +int igt_device_find_all_i915_integrated_card(struct igt_device_card **card)
> +{
> + igt_assert(card);
> + return __find_all_intel_card_by_driver_name(card, false, "i915");
> +}
> +
> +int igt_device_find_all_xe_discrete_card(struct igt_device_card **card)
> +{
> + igt_assert(card);
> + return __find_all_intel_card_by_driver_name(card, true, "xe");
> +}
> +
> +int igt_device_find_all_i915_discrete_card(struct igt_device_card **card)
> +{
> + igt_assert(card);
> + return __find_all_intel_card_by_driver_name(card, true, "i915");
> +}
> +
> static struct igt_device *igt_device_from_syspath(const char *syspath)
> {
> struct igt_device *dev;
> diff --git a/lib/igt_device_scan.h b/lib/igt_device_scan.h
> index 92741fe3c..da107292a 100644
> --- a/lib/igt_device_scan.h
> +++ b/lib/igt_device_scan.h
> @@ -59,6 +59,7 @@ struct igt_device_card {
> char subsystem[NAME_MAX];
> char card[NAME_MAX];
> char render[NAME_MAX];
> + char driver[NAME_MAX];
> char pci_slot_name[PCI_SLOT_NAME_SIZE+1];
> uint16_t pci_vendor, pci_device;
> };
> @@ -92,6 +93,10 @@ bool igt_device_find_first_i915_discrete_card(struct igt_device_card *card);
> bool igt_device_find_integrated_card(struct igt_device_card *card);
> bool igt_device_find_first_xe_discrete_card(struct igt_device_card *card);
> bool igt_device_find_xe_integrated_card(struct igt_device_card *card);
> +int igt_device_find_all_i915_discrete_card(struct igt_device_card **card);
> +int igt_device_find_all_i915_integrated_card(struct igt_device_card **card);
> +int igt_device_find_all_xe_integrated_card(struct igt_device_card **card);
> +int igt_device_find_all_xe_discrete_card(struct igt_device_card **card);
> char *igt_device_get_pretty_name(struct igt_device_card *card, bool numeric);
> int igt_open_card(struct igt_device_card *card);
> int igt_open_render(struct igt_device_card *card);
> diff --git a/lib/igt_perf.c b/lib/igt_perf.c
> index 3866c6d77..3f2f3311f 100644
> --- a/lib/igt_perf.c
> +++ b/lib/igt_perf.c
> @@ -129,6 +129,59 @@ uint64_t igt_perf_type_id(const char *device)
> return strtoull(buf, NULL, 0);
> }
>
> +int igt_perf_format(const char *device, const char *name, char *buff, int buflen)
> +{
> + char buf[NAME_MAX];
> + ssize_t ret;
> + int fd;
> +
> + snprintf(buf, sizeof(buf),
> + "/sys/bus/event_source/devices/%s/format/%s", device, name);
> +
> + fd = open(buf, O_RDONLY);
> + if (fd < 0)
> + return -1;
> +
> + ret = read(fd, buff, buflen - 1);
> + close(fd);
> + if (ret < 1)
> + return -1;
> +
> + buf[ret] = '\0';
> +
> + return 0;
> +}
> +
> +uint64_t xe_perf_event_config(int xe, const char *pmu_str)
The above 2 functions are already merged in IGT as part of my C6 tests,
you can remove them from this patch.
Thanks,
Vinay.
> +{
> + char buf[150];
> + ssize_t ret;
> + int fd;
> + uint64_t config;
> + char device[30];
> +
> + snprintf(buf, sizeof(buf),
> + "/sys/bus/event_source/devices/%s/events/%s",
> + xe_perf_device(xe, device, sizeof(device)),
> + pmu_str);
> +
> + fd = open(buf, O_RDONLY);
> + if (fd < 0)
> + return 0;
> +
> + ret = read(fd, buf, sizeof(buf) - 1);
> + close(fd);
> + if (ret < 1)
> + return 0;
> +
> + buf[ret] = '\0';
> + ret = sscanf(buf, "event=0x%lx", &config);
> + if (ret != 1)
> + return 0;
> +
> + return config;
> +}
> +
> int igt_perf_events_dir(int i915)
> {
> char buf[80];
> diff --git a/lib/igt_perf.h b/lib/igt_perf.h
> index 3d9ba2917..26b9ffa29 100644
> --- a/lib/igt_perf.h
> +++ b/lib/igt_perf.h
> @@ -54,9 +54,11 @@ perf_event_open(struct perf_event_attr *attr,
> }
>
> uint64_t igt_perf_type_id(const char *device);
> +uint64_t xe_perf_event_config(int xe, const char *pmu_event);
> int igt_perf_events_dir(int i915);
> int igt_perf_open(uint64_t type, uint64_t config);
> int igt_perf_open_group(uint64_t type, uint64_t config, int group);
> +int igt_perf_format(const char *device, const char *name, char *buff, int buflen);
>
> const char *i915_perf_device(int i915, char *buf, int buflen);
> uint64_t i915_perf_type_id(int i915);
> diff --git a/tools/gputop/common_gputop.c b/tools/gputop/common_gputop.c
> new file mode 100644
> index 000000000..1188d8e6a
> --- /dev/null
> +++ b/tools/gputop/common_gputop.c
> @@ -0,0 +1,51 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2025 Intel Corporation
> + */
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include "common_gputop.h"
> +
> +static const char * const bars[] = { " ", "▏", "▎", "▍", "▌", "▋", "▊", "▉", "█" };
> +
> +void n_spaces(const unsigned int n)
> +{
> + unsigned int i;
> +
> + for (i = 0; i < n; i++)
> + putchar(' ');
> +}
> +
> +void print_percentage_bar(double percent, int max_len)
> +{
> + int bar_len, i, len = max_len - 1;
> + const int w = 8;
> +
> + len -= printf("|%5.1f%% ", percent);
> +
> + /* no space left for bars, do what we can */
> + if (len < 0)
> + len = 0;
> +
> + bar_len = ceil(w * percent * len / 100.0);
> + if (bar_len > w * len)
> + bar_len = w * len;
> +
> + for (i = bar_len; i >= w; i -= w)
> + printf("%s", bars[w]);
> + if (i)
> + printf("%s", bars[i]);
> +
> + len -= (bar_len + (w - 1)) / w;
> + n_spaces(len);
> +
> + putchar('|');
> +}
> +
> +int print_engines_footer(int lines, int con_w, int con_h)
> +{
> + if (lines++ < con_h)
> + printf("\n");
> +
> + return lines;
> +}
> diff --git a/tools/gputop/common_gputop.h b/tools/gputop/common_gputop.h
> new file mode 100644
> index 000000000..29ba48d86
> --- /dev/null
> +++ b/tools/gputop/common_gputop.h
> @@ -0,0 +1,16 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright © 2025 Intel Corporation
> + */
> +#ifndef COMMON_GPUTOP_H
> +#define COMMON_GPUTOP_H
> +
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <math.h>
> +
> +void print_percentage_bar(double percent, int max_len);
> +int print_engines_footer(int lines, int con_w, int con_h);
> +void n_spaces(const unsigned int n);
> +
> +#endif // COMMON_GPUTOP_H
> diff --git a/tools/gputop.c b/tools/gputop/gputop.c
> similarity index 65%
> rename from tools/gputop.c
> rename to tools/gputop/gputop.c
> index 43b01f566..e53d1f087 100644
> --- a/tools/gputop.c
> +++ b/tools/gputop/gputop.c
> @@ -1,8 +1,7 @@
> // SPDX-License-Identifier: MIT
> /*
> - * Copyright © 2023 Intel Corporation
> + * Copyright © 2025 Intel Corporation
> */
> -
> #include <assert.h>
> #include <ctype.h>
> #include <dirent.h>
> @@ -31,49 +30,78 @@
> #include "igt_drm_fdinfo.h"
> #include "igt_profiling.h"
> #include "drmtest.h"
> +#include "xe/xe_query.h"
> +#include "igt_perf.h"
> +#include "igt_device_scan.h"
> +#include "xe_gputop.h"
>
> -enum utilization_type {
> - UTILIZATION_TYPE_ENGINE_TIME,
> - UTILIZATION_TYPE_TOTAL_CYCLES,
> +/*
> + * Supported Drivers
> + */
> +static const char * const drivers[] = {
> + "xe",
> +// "i915", yet to implement
> + /*Keep the last one NULL*/
> + NULL
> };
>
> -static const char *bars[] = { " ", "▏", "▎", "▍", "▌", "▋", "▊", "▉", "█" };
> -
> -#define ANSI_HEADER "\033[7m"
> -#define ANSI_RESET "\033[0m"
> +/*
> + * Number of supported drivers needs to be adjusted
> + * as per the letgth of the drivers[] array.
> + */
> +#define NUM_DRIVER 1
>
> -static void n_spaces(const unsigned int n)
> -{
> - unsigned int i;
> +/*
> + * Supported operations on driver instances.
> + * Update the array of function pointers for
> + * each individual driver specific function.
> + * Maintain the sequence as per drivers[] array.
> + */
> +void *(*discover_engines[NUM_DRIVER])(const void *obj) = {
> + xe_discover_engines
> +};
>
> - for (i = 0; i < n; i++)
> - putchar(' ');
> -}
> +void (*pmu_sample[NUM_DRIVER])(const void *obj) = {
> + xe_pmu_sample
> +};
>
> -static void print_percentage_bar(double percent, int max_len)
> -{
> - int bar_len, i, len = max_len - 1;
> - const int w = 8;
> +int (*pmu_init[NUM_DRIVER])(const void *obj) = {
> + xe_pmu_init
> +};
>
> - len -= printf("|%5.1f%% ", percent);
> +int (*print_engines[NUM_DRIVER])(const void *obj, int lines, int w, int h) = {
> + xe_print_engines
> +};
>
> - /* no space left for bars, do what we can */
> - if (len < 0)
> - len = 0;
> +/*
> + * Update this devices[] array with initialized
> + * values as per drivers[] array
> + */
> +struct gputop_device {
> + bool driver_present;
> + int len;
> + void *instances;
> +} devices[] = {
> + {false, 0, NULL}
> +};
>
> - bar_len = ceil(w * percent * len / 100.0);
> - if (bar_len > w * len)
> - bar_len = w * len;
> +enum utilization_type {
> + UTILIZATION_TYPE_ENGINE_TIME,
> + UTILIZATION_TYPE_TOTAL_CYCLES,
> +};
>
> - for (i = bar_len; i >= w; i -= w)
> - printf("%s", bars[w]);
> - if (i)
> - printf("%s", bars[i]);
> +#define ANSI_HEADER "\033[7m"
> +#define ANSI_RESET "\033[0m"
>
> - len -= (bar_len + (w - 1)) / w;
> - n_spaces(len);
> +void xe_populate_device_instances(struct gputop_device *dv);
>
> - putchar('|');
> +static int find_Driver(struct igt_device_card *card)
> +{
> + for (int i = 0; drivers[i]; i++) {
> + if (strcmp(drivers[i], card->driver) == 0)
> + return i;
> + }
> + return -1;
> }
>
> static int
> @@ -305,7 +333,6 @@ static int client_cmp(const void *_a, const void *_b, void *unused)
> return 1;
> else
> return -1;
> -
> }
>
> static void update_console_size(int *w, int *h)
> @@ -333,6 +360,7 @@ static void clrscr(void)
> struct gputop_args {
> long n_iter;
> unsigned long delay_usec;
> + char *device;
> };
>
> static void help(void)
> @@ -343,16 +371,18 @@ static void help(void)
> "\t-h, --help show this help\n"
> "\t-d, --delay =SEC[.TENTHS] iterative delay as SECS [.TENTHS]\n"
> "\t-n, --iterations =NUMBER number of executions\n"
> + "\t-D, --device Device filter"
> , program_invocation_short_name);
> }
>
> static int parse_args(int argc, char * const argv[], struct gputop_args *args)
> {
> - static const char cmdopts_s[] = "hn:d:";
> + static const char cmdopts_s[] = "hn:d:D:";
> static const struct option cmdopts[] = {
> {"help", no_argument, 0, 'h'},
> {"delay", required_argument, 0, 'd'},
> {"iterations", required_argument, 0, 'n'},
> + {"device", required_argument, 0, 'D'},
> { }
> };
>
> @@ -360,6 +390,7 @@ static int parse_args(int argc, char * const argv[], struct gputop_args *args)
> memset(args, 0, sizeof(*args));
> args->n_iter = -1;
> args->delay_usec = 2 * USEC_PER_SEC;
> + args->device = NULL;
>
> for (;;) {
> int c, idx = 0;
> @@ -383,6 +414,9 @@ static int parse_args(int argc, char * const argv[], struct gputop_args *args)
> return -1;
> }
> break;
> + case 'D':
> + args->device = optarg;
> + break;
> case 'h':
> help();
> return 0;
> @@ -403,6 +437,56 @@ static void sigint_handler(int sig)
> stop_top = true;
> }
>
> +void xe_populate_device_instances(struct gputop_device *dv)
> +{
> + struct igt_device_card *card_int = NULL, *card_dis = NULL, *cards_combi = NULL;
> + int count_int = 0, count_dis = 0;
> +
> + count_int = igt_device_find_all_xe_integrated_card(&card_int);
> + count_dis = igt_device_find_all_xe_discrete_card(&card_dis);
> +
> + if (count_int > 0 || count_dis > 0) {
> + // Allocate memory for the combined array
> + cards_combi = (struct igt_device_card *)calloc((count_int + count_dis),
> + sizeof(struct igt_device_card));
> + if (!cards_combi) {
> + fprintf(stderr, "Memory allocation failed for igt_device_card\n");
> + if (card_int)
> + free(card_int);
> + if (card_dis)
> + free(card_dis);
> + exit(EXIT_FAILURE);
> + }
> +
> + if (card_int) {
> + memcpy(cards_combi, card_int,
> + count_int * sizeof(struct igt_device_card));
> + free(card_int);
> + }
> +
> + if (card_dis) {
> + memcpy(cards_combi + count_int,
> + card_dis, count_dis * sizeof(struct igt_device_card));
> + free(card_dis);
> + }
> +
> + dv->driver_present = true;
> + dv->len = count_int + count_dis;
> + dv->instances = calloc(dv->len, sizeof(struct xe_gputop));
> + for (int i = 0; i < count_int; i++) {
> + xe_gputop_init((struct xe_gputop *)dv->instances + i,
> + cards_combi + i
> + );
> + }
> +
> + for (int i = 0; i < count_dis; i++) {
> + xe_gputop_init((struct xe_gputop *)dv->instances + count_int + i,
> + cards_combi + count_int + i
> + );
> + }
> + }
> +}
> +
> int main(int argc, char **argv)
> {
> struct gputop_args args;
> @@ -422,6 +506,85 @@ int main(int argc, char **argv)
> n = args.n_iter;
> period_us = args.delay_usec;
>
> + igt_devices_scan();
> +
> + if (args.device) {
> + struct igt_device_card *card = calloc(1, sizeof(struct igt_device_card));
> +
> + if (!igt_device_card_match(args.device, card)) {
> + printf("No device found for the filter\n"
> + "Showing for all devices\n");
> + free(card);
> + } else {
> + int driver_no = find_Driver(card);
> +
> + if (driver_no < 0) {
> + fprintf(stderr, "The driver %s could not be found.", card->driver);
> + exit(EXIT_FAILURE);
> + }
> +
> + devices[driver_no].driver_present = true;
> + devices[driver_no].len = 1;
> + switch (driver_no) {
> + case 0:
> + devices[driver_no].instances =
> + calloc(1, sizeof(struct xe_gputop));
> + xe_gputop_init(devices[driver_no].instances,
> + card
> + );
> + break;
> + }
> + goto explore_devices;
> + }
> + }
> +
> + for (int i = 0; drivers[i]; i++) {
> + switch (i) {
> + case 0: // xe
> + xe_populate_device_instances(devices + i);
> + break;
> + }
> + }
> +
> +explore_devices:
> +
> + for (int i = 0; drivers[i]; i++) {
> + if (devices[i].driver_present) {
> + for (int j = 0; j < devices[i].len; j++) {
> + if (!discover_engines[i](devices[i].instances + j)) {
> + fprintf(stderr,
> + "Failed to discover engines! (%s)\n",
> + strerror(errno));
> + return EXIT_FAILURE;
> + }
> + ret = pmu_init[i](devices[i].instances + j);
> +
> + if (ret) {
> + fprintf(stderr,
> + "Failed to initialize PMU! (%s)\n",
> + strerror(errno));
> + if (errno == EACCES && geteuid())
> + fprintf(stderr,
> + "\n"
> + "When running as a normal user CAP_PERFMON is required to access performance\n"
> + "monitoring. See \"man 7 capabilities\", \"man 8 setcap\", or contact your\n"
> + "distribution vendor for assistance.\n"
> + "\n"
> + "More information can be found at 'Perf events and tool security' document:\n"
> + "https://www.kernel.org/doc/html/latest/admin-guide/perf-security.html\n");
> +
> + igt_devices_free();
> + return EXIT_FAILURE;
> + }
> + }
> + }
> + }
> +
> + for (int i = 0; drivers[i]; i++) {
> + for (int j = 0; devices[i].driver_present && j < devices[i].len; j++)
> + pmu_sample[i](devices[i].instances + j);
> + }
> +
> clients = igt_drm_clients_init(NULL);
> if (!clients)
> exit(1);
> @@ -442,7 +605,7 @@ int main(int argc, char **argv)
>
> while ((n != 0) && !stop_top) {
> struct igt_drm_client *c, *prevc = NULL;
> - int i, engine_w = 0, lines = 0;
> + int k, engine_w = 0, lines = 0;
>
> igt_drm_clients_scan(clients, NULL, NULL, 0, NULL, 0);
> igt_drm_clients_sort(clients, client_cmp);
> @@ -450,6 +613,14 @@ int main(int argc, char **argv)
> update_console_size(&con_w, &con_h);
> clrscr();
>
> + for (int i = 0; drivers[i]; i++) {
> + for (int j = 0; devices[i].driver_present && j < devices[i].len; j++) {
> + pmu_sample[i](devices[i].instances + j);
> + lines = print_engines[i](devices[i].instances + j,
> + lines, con_w, con_h);
> + }
> + }
> +
> if (!clients->num_clients) {
> const char *msg = " (No GPU clients yet. Start workload to see stats)";
>
> @@ -457,7 +628,7 @@ int main(int argc, char **argv)
> (int)(con_w - strlen(msg) - 1), msg);
> }
>
> - igt_for_each_drm_client(clients, c, i) {
> + igt_for_each_drm_client(clients, c, k) {
> assert(c->status != IGT_DRM_CLIENT_PROBE);
> if (c->status != IGT_DRM_CLIENT_ALIVE)
> break; /* Active clients are first in the array. */
> @@ -489,3 +660,4 @@ int main(int argc, char **argv)
>
> return 0;
> }
> +
> diff --git a/tools/gputop/meson.build b/tools/gputop/meson.build
> new file mode 100644
> index 000000000..0512ac3d6
> --- /dev/null
> +++ b/tools/gputop/meson.build
> @@ -0,0 +1,6 @@
> +gputop_src = [ 'gputop.c', 'common_gputop.c', 'xe_gputop.c']
> +executable('gputop', sources : gputop_src,
> + install : true,
> + install_rpath : bindir_rpathdir,
> + dependencies : [igt_deps,lib_igt_perf,lib_igt_drm_clients,lib_igt_drm_fdinfo,lib_igt_profiling,math],
> + install: true)
> diff --git a/tools/gputop/xe_gputop.c b/tools/gputop/xe_gputop.c
> new file mode 100644
> index 000000000..2751a6e4e
> --- /dev/null
> +++ b/tools/gputop/xe_gputop.c
> @@ -0,0 +1,359 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright © 2025 Intel Corporation
> + */
> +
> +#include "xe_gputop.h"
> +#include "common_gputop.h"
> +
> +#define engine_ptr(engines, n) (&(engines)->engine + (n))
> +
> +static void __update_sample(struct xe_pmu_counter *counter, uint64_t val)
> +{
> + counter->val.prev = counter->val.cur;
> + counter->val.cur = val;
> +}
> +
> +static void update_sample(struct xe_pmu_counter *counter, uint64_t *val)
> +{
> + if (counter->present)
> + __update_sample(counter, val[counter->idx]);
> +}
> +
> +static const char *class_display_name(unsigned int class)
> +{
> + switch (class) {
> + case DRM_XE_ENGINE_CLASS_RENDER:
> + return "Render/3D";
> + case DRM_XE_ENGINE_CLASS_COPY:
> + return "Blitter";
> + case DRM_XE_ENGINE_CLASS_VIDEO_DECODE:
> + return "Video";
> + case DRM_XE_ENGINE_CLASS_VIDEO_ENHANCE:
> + return "VideoEnhance";
> + case DRM_XE_ENGINE_CLASS_COMPUTE:
> + return "Compute";
> + default:
> + return "[unknown]";
> + }
> +}
> +
> +static inline void *clean_up(void *engines)
> +{
> + if (engines)
> + free(engines);
> +
> + return NULL;
> +}
> +
> +static int _open_pmu(uint64_t type, unsigned int *cnt, struct xe_pmu_counter *pmu, int *fd)
> +{
> + int fd__ = igt_perf_open_group(type, pmu->config, *fd);
> +
> + if (fd__ >= 0) {
> + if (*fd == -1)
> + *fd = fd__;
> + pmu->present = true;
> + pmu->idx = (*cnt)++;
> + }
> +
> + return fd__;
> +}
> +
> +/* tr_pmu_name()
> + *
> + * Transliterate pci_slot_id to sysfs device name entry for discrete GPU.
> + * Discrete GPU PCI ID ("xxxx:yy:zz.z") device = "xe_xxxx_yy_zz.z".
> + */
> +static char *tr_pmu_name(const struct igt_device_card *card)
> +{
> + int ret;
> + const int bufsize = 16;
> + char *buf, *device = NULL;
> +
> + assert(card->pci_slot_name[0]);
> +
> + device = malloc(bufsize);
> + assert(device);
> +
> + ret = snprintf(device, bufsize, "xe_%s", card->pci_slot_name);
> + assert(ret == (bufsize - 1));
> +
> + buf = device;
> + for (; *buf; buf++)
> + if (*buf == ':')
> + *buf = '_';
> +
> + return device;
> +}
> +
> +void xe_gputop_init(struct xe_gputop *obj,
> + struct igt_device_card *card)
> +{
> + obj->pmu_device = tr_pmu_name(card);
> + obj->card = card;
> +}
> +
> +static int pmu_format_shift(int xe, const char *name)
> +{
> + int start, end, ret;
> + int format;
> + char device[80], buff[80];
> +
> + format = igt_perf_format(xe_perf_device(xe, device, sizeof(device)),
> + name, buff, sizeof(buff));
> + if (format)
> + return 0;
> +
> + ret = sscanf(buff, "config:%d-%d", &start, &end);
> + igt_assert(ret >= 1);
> +
> + return start;
> +}
> +
> +static int engine_cmp(const void *__a, const void *__b)
> +{
> + const struct xe_engine *a = (struct xe_engine *)__a;
> + const struct xe_engine *b = (struct xe_engine *)__b;
> +
> + if (a->drm_xe_engine.engine_class != b->drm_xe_engine.engine_class)
> + return a->drm_xe_engine.engine_class - b->drm_xe_engine.engine_class;
> + else
> + return a->drm_xe_engine.engine_instance - b->drm_xe_engine.engine_instance;
> +}
> +
> +void *xe_discover_engines(const void *obj)
> +{
> + struct igt_device_card *card = ((struct xe_gputop *)obj)->card;
> + struct xe_engines *engines;
> + int ret = 0;
> + DIR *d;
> + struct drm_xe_engine_class_instance *hwe;
> + int card_fd;
> +
> + if (!card || !strlen(card->card) || !strlen(card->render))
> + return NULL;
> +
> + if (strlen(card->card)) {
> + card_fd = igt_open_card(card);
> + } else if (strlen(card->render)) {
> + card_fd = igt_open_render(card);
> + } else {
> + fprintf(stderr, "Failed to detect device!\n");
> + return clean_up(engines);
> + }
> + xe_device_get(card_fd);
> + engines = malloc(sizeof(struct xe_engines));
> + if (!engines)
> + return NULL;
> +
> + memset(engines, 0, sizeof(*xe_engines));
> +
> + engines->num_engines = 0;
> + engines->device = ((struct xe_gputop *)obj)->pmu_device;
> + xe_for_each_engine(card_fd, hwe) {
> + uint64_t engine_class, engine_instance, gt_shift, param_config;
> + struct xe_engine *engine;
> +
> + engine = engine_ptr(engines, engines->num_engines);
> + gt_shift = pmu_format_shift(card_fd, "gt");
> + engine_class = pmu_format_shift(card_fd, "engine_class");
> + engine_instance = pmu_format_shift(card_fd, "engine_instance");
> + param_config = (uint64_t)hwe->gt_id << gt_shift | hwe->engine_class << engine_class
> + | hwe->engine_instance << engine_instance;
> +
> + engine->drm_xe_engine = *hwe;
> + engine->busy.config = xe_perf_event_config(card_fd, "engine-active-ticks")
> + | param_config;
> + engine->total.config = xe_perf_event_config(card_fd, "engine-total-ticks")
> + | param_config;
> +
> + if (engine->busy.config == -1 || engine->total.config == -1) {
> + ret = ENOENT;
> + break;
> + }
> +
> + ret = asprintf(&engine->display_name, "%s/%u",
> + class_display_name(engine->drm_xe_engine.engine_class),
> + engine->drm_xe_engine.engine_instance);
> +
> + if (ret <= 0) {
> + ret = errno;
> + break;
> + }
> + ret = asprintf(&engine->short_name, "%s/%u",
> + xe_engine_class_short_string(engine->drm_xe_engine.engine_class),
> + engine->drm_xe_engine.engine_instance);
> +
> + if (ret <= 0) {
> + ret = errno;
> + break;
> + }
> +
> + engines->num_engines++;
> + engines = realloc(engines, sizeof(struct xe_engines) +
> + engines->num_engines * sizeof(struct xe_engine));
> + if (!engines) {
> + ret = errno;
> + break;
> + }
> + }
> +
> + if (!ret) {
> + errno = ret;
> + return clean_up(engines);
> + }
> +
> + qsort(engine_ptr(engines, 0), engines->num_engines,
> + sizeof(struct xe_engine), engine_cmp);
> +
> + engines->root = d;
> + ((struct xe_gputop *)obj)->eng_obj = engines;
> +
> + return engines;
> +}
> +
> +static uint64_t pmu_read_multi(int fd, unsigned int num, uint64_t *val)
> +{
> + uint64_t buf[2 + num];
> + unsigned int i;
> + ssize_t len;
> +
> + memset(buf, 0, sizeof(buf));
> +
> + len = read(fd, buf, sizeof(buf));
> + assert(len == sizeof(buf));
> +
> + for (i = 0; i < num; i++)
> + val[i] = buf[2 + i];
> +
> + return buf[1];
> +}
> +
> +void xe_pmu_sample(const void *obj)
> +{
> + struct xe_engines *engines = ((struct xe_gputop *)obj)->eng_obj;
> + const int num_val = engines->num_counters;
> + uint64_t val[2 + num_val];
> + unsigned int i;
> +
> + engines->ts.prev = engines->ts.cur;
> + engines->ts.cur = pmu_read_multi(engines->fd, num_val, val);
> +
> + for (i = 0; i < engines->num_engines; i++) {
> + struct xe_engine *engine = engine_ptr(engines, i);
> +
> + update_sample(&engine->busy, val);
> + update_sample(&engine->total, val);
> + }
> +}
> +
> +int xe_pmu_init(const void *obj)
> +{
> + struct xe_engines *engines = ((struct xe_gputop *)obj)->eng_obj;
> + unsigned int i;
> + int fd;
> + struct xe_engine *engine;
> + uint64_t type = igt_perf_type_id(engines->device);
> +
> + engines->fd = -1;
> + engines->num_counters = 0;
> +
> + engine = engine_ptr(engines, 0);
> + fd = _open_pmu(type, &engines->num_counters, &engine->busy, &engines->fd);
> + if (fd < 0)
> + return -1;
> + fd = _open_pmu(type, &engines->num_counters, &engine->total, &engines->fd);
> + if (fd < 0)
> + return -1;
> +
> + for (i = 1; i < engines->num_engines; i++) {
> + engine = engine_ptr(engines, i);
> + fd = _open_pmu(type, &engines->num_counters, &engine->busy, &engines->fd);
> + if (fd < 0)
> + return -1;
> + fd = _open_pmu(type, &engines->num_counters, &engine->total, &engines->fd);
> + if (fd < 0)
> + return -1;
> + }
> + return 0;
> +}
> +
> +static double pmu_calc_total(struct xe_pmu_pair *p)
> +{
> + double v;
> +
> + v = (p->cur - p->prev) / 1e9;
> + return v;
> +}
> +
> +static double pmu_calc(struct xe_pmu_pair *p, double total_tick)
> +{
> + double bz = (p->cur - p->prev) / 1e9;
> + double total;
> +
> + total = (bz * 100) / total_tick;
> + return total;
> +}
> +
> +static int
> +print_engines_header(struct xe_engines *engines,
> + int lines, int con_w, int con_h)
> +{
> + const char *a;
> +
> + for (unsigned int i = 0;
> + i < engines->num_engines && lines < con_h;
> + i++) {
> + struct xe_engine *engine = engine_ptr(engines, i);
> +
> + if (!engine->num_counters)
> + continue;
> +
> + a = " ENGINES BUSY ";
> +
> + printf("\033[7m%s%*s\033[0m\n",
> + a,
> + (int)(con_w - strlen(a)), " ");
> +
> + lines++;
> +
> + break;
> + }
> +
> + return lines;
> +}
> +
> +static int
> +print_engine(struct xe_engines *engines, unsigned int i,
> + int lines, int con_w, int con_h)
> +{
> + struct xe_engine *engine = engine_ptr(engines, i);
> + double total_tick = pmu_calc_total(&engine->total.val);
> + double percentage = pmu_calc(&engine->busy.val, total_tick);
> +
> + printf("%*s", (int)(strlen(" ENGINES")), engine->display_name);
> + //printf(" %5.1f", percentage);
> + print_percentage_bar(percentage, con_w - strlen(" ENGINES"));
> + printf("\n");
> +
> + return ++lines;
> +}
> +
> +int xe_print_engines(const void *obj, int lines, int w, int h)
> +{
> + struct xe_engines *engines = ((struct xe_gputop *)obj)->eng_obj;
> + struct xe_engines *show;
> +
> + show = engines;
> +
> + lines = print_engines_header(show, lines, w, h);
> +
> + for (unsigned int i = 0; i < show->num_engines && lines < h; i++)
> + lines = print_engine(show, i, lines, w, h);
> +
> + lines = print_engines_footer(lines, w, h);
> +
> + return lines;
> +}
> +
> diff --git a/tools/gputop/xe_gputop.h b/tools/gputop/xe_gputop.h
> new file mode 100644
> index 000000000..0f7291563
> --- /dev/null
> +++ b/tools/gputop/xe_gputop.h
> @@ -0,0 +1,74 @@
> +/* SPDX-License-Identifier: MIT
> + *
> + * Copyright © 2025 Intel Corporation
> + */
> +
> +#ifndef __XE_GPUTOP_H__
> +#define __XE_GPUTOP_H__
> +
> +#include <dirent.h>
> +
> +#include "igt_device_scan.h"
> +#include "xe/xe_query.h"
> +#include "igt_perf.h"
> +#include "common_gputop.h"
> +
> +struct xe_pmu_pair {
> + uint64_t cur;
> + uint64_t prev;
> +};
> +
> +struct xe_pmu_counter {
> + uint64_t type;
> + uint64_t config;
> + unsigned int idx;
> + struct xe_pmu_pair val;
> + bool present;
> +};
> +
> +struct xe_engine {
> + const char *name;
> + char *display_name;
> + char *short_name;
> + struct drm_xe_engine_class_instance drm_xe_engine;
> + unsigned int num_counters;
> + struct xe_pmu_counter busy;
> + struct xe_pmu_counter total;
> +};
> +
> +struct xe_engines {
> + unsigned int num_engines;
> + unsigned int num_classes;
> + unsigned int num_counters;
> + DIR *root;
> + int fd;
> + struct xe_pmu_pair ts;
> + bool discrete;
> + char *device;
> + int num_gts;
> +
> + /* Do not edit below this line.
> + * This structure is reallocated every time a new engine is
> + * found and size is increased by sizeof (engine).
> + */
> +
> + struct xe_engine engine;
> +
> +};
> +
> +struct xe_gputop {
> + char *pmu_device;
> + struct igt_device_card *card;
> + struct xe_engines *eng_obj;
> +};
> +
> +void xe_gputop_init(struct xe_gputop *obj,
> + struct igt_device_card *card);
> +
> +void *xe_discover_engines(const void *obj);
> +void xe_pmu_sample(const void *obj);
> +int xe_pmu_init(const void *obj);
> +int xe_print_engines(const void *obj, int lines, int w, int h);
> +
> +#endif // __XE_GPUTOP_H__
> +
> diff --git a/tools/meson.build b/tools/meson.build
> index f091af380..7a9fdfb9c 100644
> --- a/tools/meson.build
> +++ b/tools/meson.build
> @@ -68,11 +68,6 @@ if libudev.found()
> install : true)
> endif
>
> -executable('gputop', 'gputop.c',
> - install : true,
> - install_rpath : bindir_rpathdir,
> - dependencies : [lib_igt_drm_clients,lib_igt_drm_fdinfo,lib_igt_profiling,math])
> -
> intel_l3_parity_src = [ 'intel_l3_parity.c', 'intel_l3_udev_listener.c' ]
> executable('intel_l3_parity', sources : intel_l3_parity_src,
> dependencies : tool_deps,
> @@ -121,3 +116,4 @@ endif
> subdir('i915-perf')
> subdir('xe-perf')
> subdir('null_state_gen')
> +subdir('gputop')
> \ No newline at end of file
More information about the igt-dev
mailing list