From: Mario Limonciello <mario.limonciello@amd.com>
To: Pratap Nirujogi <pratap.nirujogi@amd.com>,
amd-gfx@lists.freedesktop.org, mlimonci@amd.com,
alexander.deucher@amd.com, christian.koenig@amd.com
Cc: benjamin.chan@amd.com, bin.du@amd.com,
gjorgji.rosikopulos@amd.com, king.li@amd.com, dantony@amd.com,
phil.jawich@amd.com, Gjorgji Rosikopulos <grosikop@amd.com>
Subject: Re: [PATCH] drm/amd/amdgpu: Fix SMU warning during isp suspend-resume
Date: Tue, 9 Dec 2025 21:28:36 -0600 [thread overview]
Message-ID: <db479aed-d96d-414c-88f2-a699accf24f8@amd.com> (raw)
In-Reply-To: <20251210015140.568540-1-pratap.nirujogi@amd.com>
On 12/9/2025 7:50 PM, Pratap Nirujogi wrote:
> ISP mfd child devices are using genpd and the system suspend-resume
> operations between genpd and amdgpu parent device which uses only
> runtime suspend-resume are not in sync.
>
> Linux power manager during suspend-resume resuming the genpd devices
> earlier than the amdgpu parent device. This is resulting in the below
> warning as SMU is in suspended state when genpd attempts to resume ISP.
>
> WARNING: CPU: 13 PID: 5435 at drivers/gpu/drm/amd/amdgpu/../pm/swsmu/amdgpu_smu.c:398 smu_dpm_set_power_gate+0x36f/0x380 [amdgpu]
>
> To fix this warning isp suspend-resume is handled as part of amdgpu
> parent device suspend-resume instead of genpd sequence. Each ISP MFD
> child device is marked as dev_pm_syscore_device to skip genpd
> suspend-resume and use pm_runtime_force api's to suspend-resume
> the devices when callbacks from amdgpu are received.
>
> Signed-off-by: Gjorgji Rosikopulos <grosikop@amd.com>
> Signed-off-by: Bin Du <bin.du@amd.com>
> Signed-off-by: Pratap Nirujogi <pratap.nirujogi@amd.com>
Who is the patch author? If you guys worked together, there should be
Co-developed-by tags to represent it.
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c | 24 ++++++++++
> drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h | 2 +
> drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c | 59 +++++++++++++++++++++++++
> 3 files changed, 85 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
> index 37270c4dab8d..532f83d783d1 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c
> @@ -318,12 +318,36 @@ void isp_kernel_buffer_free(void **buf_obj, u64 *gpu_addr, void **cpu_addr)
> }
> EXPORT_SYMBOL(isp_kernel_buffer_free);
>
> +static int isp_resume(struct amdgpu_ip_block *ip_block)
> +{
> + struct amdgpu_device *adev = ip_block->adev;
> + struct amdgpu_isp *isp = &adev->isp;
> +
> + if (isp->funcs->hw_resume)
> + return isp->funcs->hw_resume(isp);
> +
> + return -ENODEV;
> +}
> +
> +static int isp_suspend(struct amdgpu_ip_block *ip_block)
> +{
> + struct amdgpu_device *adev = ip_block->adev;
> + struct amdgpu_isp *isp = &adev->isp;
> +
> + if (isp->funcs->hw_suspend)
> + return isp->funcs->hw_suspend(isp);
> +
> + return -ENODEV;
> +}
> +
> static const struct amd_ip_funcs isp_ip_funcs = {
> .name = "isp_ip",
> .early_init = isp_early_init,
> .hw_init = isp_hw_init,
> .hw_fini = isp_hw_fini,
> .is_idle = isp_is_idle,
> + .suspend = isp_suspend,
> + .resume = isp_resume,
> .set_clockgating_state = isp_set_clockgating_state,
> .set_powergating_state = isp_set_powergating_state,
> };
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
> index d6f4ffa4c97c..9a5d2b1dff9e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.h
> @@ -38,6 +38,8 @@ struct amdgpu_isp;
> struct isp_funcs {
> int (*hw_init)(struct amdgpu_isp *isp);
> int (*hw_fini)(struct amdgpu_isp *isp);
> + int (*hw_suspend)(struct amdgpu_isp *isp);
> + int (*hw_resume)(struct amdgpu_isp *isp);
> };
>
> struct amdgpu_isp {
> diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
> index 4258d3e0b706..560c398e14fc 100644
> --- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
> +++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c
> @@ -26,6 +26,7 @@
> */
>
> #include <linux/gpio/machine.h>
> +#include <linux/pm_runtime.h>
> #include "amdgpu.h"
> #include "isp_v4_1_1.h"
>
> @@ -145,6 +146,9 @@ static int isp_genpd_add_device(struct device *dev, void *data)
> return -ENODEV;
> }
>
> + /* The devcies will be managed by the pm ops from the parent */
devices
> + dev_pm_syscore_device(dev, true);
> +
> exit:
> /* Continue to add */
> return 0;
> @@ -177,12 +181,65 @@ static int isp_genpd_remove_device(struct device *dev, void *data)
> drm_err(&adev->ddev, "Failed to remove dev from genpd %d\n", ret);
> return -ENODEV;
> }
> + dev_pm_syscore_device(dev, false);
>
> exit:
> /* Continue to remove */
> return 0;
> }
>
> +static int isp_suspend_device(struct device *dev, void *data)
> +{
> + struct platform_device *pdev = container_of(dev, struct platform_device, dev);
> +
> + if (!dev->type || !dev->type->name)
> + return 0;
> + if (strncmp(dev->type->name, "mfd_device", 10))
> + return 0;
> + if (!strncmp(pdev->mfd_cell->name, "amdisp-pinctrl", 14))
> + return 0;
Could we store the mfd_cell pointer instead and just compare the pointers?
> +
> + return pm_runtime_force_suspend(dev);
> +}
> +
> +static int isp_resume_device(struct device *dev, void *data)
> +{
> + struct platform_device *pdev = container_of(dev, struct platform_device, dev);
> +
> + if (!dev->type || !dev->type->name)
> + return 0;
> + if (strncmp(dev->type->name, "mfd_device", 10))
> + return 0;
> + if (!strncmp(pdev->mfd_cell->name, "amdisp-pinctrl", 14))
> + return 0;
same comment as above
> +
> + return pm_runtime_force_resume(dev);
> +}
> +
> +static int isp_v4_1_1_hw_suspend(struct amdgpu_isp *isp)
> +{
> + int r;
> +
> + r = device_for_each_child(isp->parent, NULL,
> + isp_suspend_device);
> + if (r)
> + dev_err(isp->parent, "failed to suspend hw devices (%d)\n", r);
> +
> + return 0;
Shouldn't you return r?
> +}
> +
> +static int isp_v4_1_1_hw_resume(struct amdgpu_isp *isp)
> +{
> + int r;
> +
> + r = device_for_each_child(isp->parent, NULL,
> + isp_resume_device);
> + if (r)
> + dev_err(isp->parent, "failed to resume hw device (%d)\n", r);
> +
> + return 0;
Shouldn't you return r?
> +}
> +
> static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp)
> {
> const struct software_node *amd_camera_node, *isp4_node;
> @@ -369,6 +426,8 @@ static int isp_v4_1_1_hw_fini(struct amdgpu_isp *isp)
> static const struct isp_funcs isp_v4_1_1_funcs = {
> .hw_init = isp_v4_1_1_hw_init,
> .hw_fini = isp_v4_1_1_hw_fini,
> + .hw_suspend = isp_v4_1_1_hw_suspend,
> + .hw_resume = isp_v4_1_1_hw_resume,
> };
>
> void isp_v4_1_1_set_isp_funcs(struct amdgpu_isp *isp)
next prev parent reply other threads:[~2025-12-10 3:28 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-12-10 1:50 [PATCH] drm/amd/amdgpu: Fix SMU warning during isp suspend-resume Pratap Nirujogi
2025-12-10 3:28 ` Mario Limonciello [this message]
2025-12-10 15:27 ` Nirujogi, Pratap
2025-12-10 23:13 ` Nirujogi, Pratap
2025-12-11 2:19 ` Mario Limonciello
2025-12-11 4:01 ` Nirujogi, Pratap
2025-12-11 18:44 ` Nirujogi, Pratap
2025-12-11 14:33 ` Alex Deucher
2025-12-11 19:31 ` Nirujogi, Pratap
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=db479aed-d96d-414c-88f2-a699accf24f8@amd.com \
--to=mario.limonciello@amd.com \
--cc=alexander.deucher@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=benjamin.chan@amd.com \
--cc=bin.du@amd.com \
--cc=christian.koenig@amd.com \
--cc=dantony@amd.com \
--cc=gjorgji.rosikopulos@amd.com \
--cc=grosikop@amd.com \
--cc=king.li@amd.com \
--cc=mlimonci@amd.com \
--cc=phil.jawich@amd.com \
--cc=pratap.nirujogi@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