From: Michael Strawbridge <michael.strawbridge@amd.com>
To: vitaly.prosyak@amd.com, igt-dev@lists.freedesktop.org
Cc: Christian Koenig <christian.koenig@amd.com>,
Alexander Deucher <alexander.deucher@amd.com>,
Jesse Zhang <jesse.zhang@amd.com>
Subject: Re: [PATCH] amd/lib: refactor function is_reset_enable
Date: Tue, 28 Jan 2025 15:43:54 -0500 [thread overview]
Message-ID: <49ab2d75-3a6a-4e60-9271-f4049bc1de20@amd.com> (raw)
In-Reply-To: <20250128192442.61513-1-vitaly.prosyak@amd.com>
Hi Vitaly,
One small thing below.
On 1/28/25 14:24, vitaly.prosyak@amd.com wrote:
> From: Vitaly Prosyak <vitaly.prosyak@amd.com>
>
> This function takes a reset type mask (`AMDGPU_RESET_TYPE_FULL`,
> `AMDGPU_RESET_TYPE_SOFT_RESET`, AMDGPU_RESET_TYPE_PER_PIPE
> or `AMDGPU_RESET_TYPE_PER_QUEUE`) and attempts to open and
> read the corresponding sysfs file, e.g.,
> `/sys/bus/pci/devices/0000:08:00.0/gfx_reset_mask`.
> It then iterates through predefined keywords such as `full`,
> `soft`, and `queue`, validating whether AMDGPU exposes the
> requested operation. If the requested operation is not available,
> the test is skipped.
>
> The previous implementation included redundant steps and failed
> in certain cases. This refactored version simplifies the logic,
> improves reliability, and ensures consistent behavior across various
> scenarios.
>
> Cc: Christian Koenig <christian.koenig@amd.com>
> Cc: Alexander Deucher <alexander.deucher@amd.com>
> Cc: Jesse Zhang <jesse.zhang@amd.com>
> Cc: Michael Strawbridge <michael.strawbridge@amd.com>
>
> Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
> ---
> lib/amdgpu/amd_ip_blocks.c | 79 +++++++++++++++++++++++++-------------
> 1 file changed, 53 insertions(+), 26 deletions(-)
>
> diff --git a/lib/amdgpu/amd_ip_blocks.c b/lib/amdgpu/amd_ip_blocks.c
> index 2b1b84414..f063ad0ab 100644
> --- a/lib/amdgpu/amd_ip_blocks.c
> +++ b/lib/amdgpu/amd_ip_blocks.c
> @@ -1006,11 +1006,27 @@ asic_rings_readness(amdgpu_device_handle device_handle, uint32_t mask,
> bool
> is_reset_enable(enum amd_ip_block_type ip_type, uint32_t reset_type, const struct pci_addr *pci)
> {
> - char cmd[256];
> - FILE *fp, *fp2;
> - char buffer[100];
> + const struct reset_arr {
> + const char *name;
> + unsigned int reset;
> + } reset_arr[] = {
> + {"full", AMDGPU_RESET_TYPE_FULL },
> + {"soft", AMDGPU_RESET_TYPE_SOFT_RESET},
> + {"queue", AMDGPU_RESET_TYPE_PER_QUEUE },
> + {"pipe", AMDGPU_RESET_TYPE_PER_PIPE },
> + {NULL, 0}
> + };
> +
> bool enable = false;
> - char reset_mask[100];
> + char cmd[256];
> + FILE *fp;
> + char buffer[128];
> +
> + char *token, *buf;
> + char reset_mask[32];
> + uint32_t mask = 0;
> + int found = 0;
> + int i;
>
> if (ip_type == AMD_IP_GFX)
> snprintf(reset_mask, sizeof(reset_mask) - 1, "gfx_reset_mask");
> @@ -1019,33 +1035,44 @@ is_reset_enable(enum amd_ip_block_type ip_type, uint32_t reset_type, const struc
> else
> snprintf(reset_mask, sizeof(reset_mask) - 1, "sdma_reset_mask");
>
> - if( pci)
> - snprintf(cmd, sizeof(cmd) - 1, "sudo cat /sys/kernel/debug/dri/%04x:%02x:%02x.%01x/name |grep -oP '(?<=dev=)[0-9:.]+'",
> - pci->domain, pci->bus, pci->device, pci->function);
> - else
> - snprintf(cmd, sizeof(cmd) - 1, "sudo cat /sys/kernel/debug/dri/0/name |grep -oP '(?<=dev=)[0-9:.]+'");
> + snprintf(cmd, sizeof(cmd) - 1, "sudo cat /sys/bus/pci/devices/%04x:%02x:%02x.%01x/%s",
> + pci->domain, pci->bus, pci->device, pci->function, reset_mask);
>
> fp = popen(cmd, "r");
> - if (fp == NULL)
> + if (fp == NULL) {
> + igt_kmsg("***FAILURE popen %s LINE %d FILE %s\n", cmd, __LINE__, __FILE__);
> return false;
> + }
>
> - if (fgets(buffer, 13, fp) != NULL) {
> - snprintf(cmd, sizeof(cmd) - 1, "sudo cat /sys/bus/pci/devices/%s/%s | grep -oP '%s'",
> - buffer, reset_mask,
> - reset_type & AMDGPU_RESET_TYPE_FULL ? "full" :
> - reset_type & AMDGPU_RESET_TYPE_SOFT_RESET ? "soft" :
> - reset_type & AMDGPU_RESET_TYPE_PER_QUEUE ? "queue" : "pipe");
> -
> - fp2 = popen(cmd, "r");
> - if (fp2 == NULL) {
> - pclose(fp);
> - return false;
> + buf = fgets(buffer, sizeof(buffer)-1, fp);
> +
> + //igt_kmsg("cmd: %s buf: %s LINE %d FILE %s\n",
> + // cmd, buf != NULL ? buf : "NULL", __LINE__, __FILE__);
Did you mean to keep the 2 above commented lines?
Besides this you can add:
Reviewed-by: Michael Strawbridge <michael.strawbridge@amd.com>
> +
> + if (buf != NULL) {
> + token = strtok(buf, " \n");
> + while (token != NULL) {
> + for (i = 0; reset_arr[i].name != NULL; i++) {
> + if (reset_type == reset_arr[i].reset &&
> + strcmp(token, reset_arr[i].name) == 0) {
> + mask |= reset_arr[i].reset;
> + found = 1;
> + break;
> + }
> + }
> + if (!found)
> + token = strtok(NULL, " \n");
> }
> -
> - if (fgets(buffer, 13, fp2) != NULL)
> - enable = true;
> - pclose(fp2);
> + } else {
> + igt_kmsg("***FAILURE fgets %s LINE %d FILE %s\n",
> + buffer, __LINE__, __FILE__);
> }
> + if (mask == reset_type)
> + enable = true;
> + else
> + igt_kmsg("***FAILURE mask found 0x%x requested 0x%x operation is not supported LINE %d FILE %s\n",
> + mask, reset_type, __LINE__, __FILE__);
> +
> pclose(fp);
>
> return enable;
> @@ -1105,7 +1132,7 @@ int get_pci_addr_from_fd(int fd, struct pci_addr *pci)
>
> if (ret != 4) {
> igt_info("error %s Unable to extract PCI device address from '%s\n",
> - __FUNCTION__, buf);
> + __func__, buf);
> return -ENOENT;
> }
>
next prev parent reply other threads:[~2025-01-28 20:44 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-01-28 19:24 [PATCH] amd/lib: refactor function is_reset_enable vitaly.prosyak
2025-01-28 20:43 ` Michael Strawbridge [this message]
2025-01-28 20:48 ` vitaly prosyak
2025-01-28 21:39 ` ✓ Xe.CI.BAT: success for amd/lib: refactor function is_reset_enable (rev2) Patchwork
2025-01-28 21:42 ` ✓ i915.CI.BAT: " Patchwork
2025-01-29 12:10 ` ✗ i915.CI.Full: failure " Patchwork
2025-01-29 12:11 ` ✗ Xe.CI.Full: " Patchwork
-- strict thread matches above, loose matches on Subject: below --
2025-01-28 1:00 [PATCH] amd/lib: refactor function is_reset_enable vitaly.prosyak
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=49ab2d75-3a6a-4e60-9271-f4049bc1de20@amd.com \
--to=michael.strawbridge@amd.com \
--cc=alexander.deucher@amd.com \
--cc=christian.koenig@amd.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=jesse.zhang@amd.com \
--cc=vitaly.prosyak@amd.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox