From: zhoucm1 <david1.zhou-5C7GfCeVMHo@public.gmane.org>
To: "Christian König"
<deathsimple-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>,
"Deucher,
Alexander" <Alexander.Deucher-5C7GfCeVMHo@public.gmane.org>,
"amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org"
<amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org>
Subject: Re: [PATCH 2/4] drm/amdgpu: return -ENODEV to user space when vram is lost v2
Date: Wed, 24 May 2017 10:20:46 +0800 [thread overview]
Message-ID: <5924EDFE.4010102@amd.com> (raw)
In-Reply-To: <69717c0b-b2c1-589a-c466-5d6be9518eda-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
On 2017年05月23日 23:16, Christian König wrote:
> Am 23.05.2017 um 17:08 schrieb Deucher, Alexander:
>>> -----Original Message-----
>>> From: amd-gfx [mailto:amd-gfx-bounces@lists.freedesktop.org] On Behalf
>>> Of Chunming Zhou
>>> Sent: Tuesday, May 16, 2017 5:26 AM
>>> To: amd-gfx@lists.freedesktop.org
>>> Cc: Zhou, David(ChunMing)
>>> Subject: [PATCH 2/4] drm/amdgpu: return -ENODEV to user space when
>>> vram is lost v2
>>>
>>> below ioctl will return -ENODEV:
>>> amdgpu_cs_ioctl
>>> amdgpu_cs_wait_ioctl
>>> amdgpu_cs_wait_fences_ioctl
>>> amdgpu_gem_va_ioctl
>>> amdgpu_info_ioctl
>> Do we want to block the info ioctl? Isn't that where the lost
>> context query is?
>
> No, that's amdgpu_ctx_ioctl.
>
> But I think the conclusion is that we want to move the vram_lost
> counter to be per CTX and not per device.
Yes, Monk is working on it for virt case, after it, I think we can reuse it.
Regards,
David zhou
>
> Christian.
>
>>
>> Alex
>>
>>> v2: only for map and replace cases in amdgpu_gem_va_ioctl
>>>
>>> Change-Id: I8970cde3301b7cfeb4263cc0f0e54aece215c98e
>>> Signed-off-by: Chunming Zhou <David1.Zhou@amd.com>
>>> ---
>>> drivers/gpu/drm/amd/amdgpu/amdgpu.h | 4 ++++
>>> drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 9 +++++++++
>>> drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 +++-
>>> drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 5 +++++
>>> drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 10 ++++++++++
>>> 5 files changed, 31 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>> index f9da215..dcd6203 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>>> @@ -855,6 +855,7 @@ struct amdgpu_fpriv {
>>> struct amdgpu_ctx_mgr ctx_mgr;
>>> spinlock_t sem_handles_lock;
>>> struct idr sem_handles;
>>> + u32 vram_lost_counter;
>>> };
>>>
>>> /*
>>> @@ -1607,6 +1608,7 @@ struct amdgpu_device {
>>> atomic64_t num_bytes_moved;
>>> atomic64_t num_evictions;
>>> atomic_t gpu_reset_counter;
>>> + atomic_t vram_lost_counter;
>>>
>>> /* data for buffer migration throttling */
>>> struct {
>>> @@ -2005,6 +2007,8 @@ static inline void
>>> amdgpu_unregister_atpx_handler(void) {}
>>> extern const struct drm_ioctl_desc amdgpu_ioctls_kms[];
>>> extern const int amdgpu_max_kms_ioctl;
>>>
>>> +bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
>>> + struct amdgpu_fpriv *fpriv);
>>> int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long
>>> flags);
>>> int amdgpu_driver_unload_kms(struct drm_device *dev);
>>> void amdgpu_driver_lastclose_kms(struct drm_device *dev);
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>>> index b803412..911aa02 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
>>> @@ -1097,6 +1097,7 @@ static int amdgpu_cs_submit(struct
>>> amdgpu_cs_parser *p,
>>> int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct
>>> drm_file
>>> *filp)
>>> {
>>> struct amdgpu_device *adev = dev->dev_private;
>>> + struct amdgpu_fpriv *fpriv = filp->driver_priv;
>>> union drm_amdgpu_cs *cs = data;
>>> struct amdgpu_cs_parser parser = {};
>>> bool reserved_buffers = false;
>>> @@ -1104,6 +1105,8 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void
>>> *data, struct drm_file *filp)
>>>
>>> if (!adev->accel_working)
>>> return -EBUSY;
>>> + if (amdgpu_kms_vram_lost(adev, fpriv))
>>> + return -ENODEV;
>>>
>>> parser.adev = adev;
>>> parser.filp = filp;
>>> @@ -1165,12 +1168,15 @@ int amdgpu_cs_wait_ioctl(struct drm_device
>>> *dev, void *data,
>>> {
>>> union drm_amdgpu_wait_cs *wait = data;
>>> struct amdgpu_device *adev = dev->dev_private;
>>> + struct amdgpu_fpriv *fpriv = filp->driver_priv;
>>> unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout);
>>> struct amdgpu_ring *ring = NULL;
>>> struct amdgpu_ctx *ctx;
>>> struct fence *fence;
>>> long r;
>>>
>>> + if (amdgpu_kms_vram_lost(adev, fpriv))
>>> + return -ENODEV;
>>> r = amdgpu_cs_get_ring(adev, wait->in.ip_type, wait-
>>>> in.ip_instance,
>>> wait->in.ring, &ring);
>>> if (r)
>>> @@ -1344,12 +1350,15 @@ int amdgpu_cs_wait_fences_ioctl(struct
>>> drm_device *dev, void *data,
>>> struct drm_file *filp)
>>> {
>>> struct amdgpu_device *adev = dev->dev_private;
>>> + struct amdgpu_fpriv *fpriv = filp->driver_priv;
>>> union drm_amdgpu_wait_fences *wait = data;
>>> uint32_t fence_count = wait->in.fence_count;
>>> struct drm_amdgpu_fence *fences_user;
>>> struct drm_amdgpu_fence *fences;
>>> int r;
>>>
>>> + if (amdgpu_kms_vram_lost(adev, fpriv))
>>> + return -ENODEV;
>>> /* Get the fences from userspace */
>>> fences = kmalloc_array(fence_count, sizeof(struct
>>> drm_amdgpu_fence),
>>> GFP_KERNEL);
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>> index c56ae4a..2f0fcf8 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
>>> @@ -2913,8 +2913,10 @@ int amdgpu_gpu_reset(struct amdgpu_device
>>> *adev)
>>> if (r)
>>> goto out;
>>> vram_lost = amdgpu_check_vram_lost(adev);
>>> - if (vram_lost)
>>> + if (vram_lost) {
>>> DRM_ERROR("VRAM is lost!\n");
>>> + atomic_inc(&adev->vram_lost_counter);
>>> + }
>>> r = amdgpu_ttm_recover_gart(adev);
>>> if (r)
>>> goto out;
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>> index d8275ef..83bc94c 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>>> @@ -802,6 +802,11 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev,
>>> void *data,
>>> args->operation);
>>> return -EINVAL;
>>> }
>>> + if ((args->operation == AMDGPU_VA_OP_MAP) ||
>>> + (args->operation == AMDGPU_VA_OP_REPLACE)) {
>>> + if (amdgpu_kms_vram_lost(adev, fpriv))
>>> + return -ENODEV;
>>> + }
>>>
>>> INIT_LIST_HEAD(&list);
>>> if ((args->operation != AMDGPU_VA_OP_CLEAR) &&
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
>>> index 368829a..a231aa1 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
>>> @@ -235,6 +235,7 @@ static int amdgpu_firmware_info(struct
>>> drm_amdgpu_info_firmware *fw_info,
>>> static int amdgpu_info_ioctl(struct drm_device *dev, void *data,
>>> struct
>>> drm_file *filp)
>>> {
>>> struct amdgpu_device *adev = dev->dev_private;
>>> + struct amdgpu_fpriv *fpriv = filp->driver_priv;
>>> struct drm_amdgpu_info *info = data;
>>> struct amdgpu_mode_info *minfo = &adev->mode_info;
>>> void __user *out = (void __user
>>> *)(uintptr_t)info->return_pointer;
>>> @@ -247,6 +248,8 @@ static int amdgpu_info_ioctl(struct drm_device
>>> *dev,
>>> void *data, struct drm_file
>>>
>>> if (!info->return_size || !info->return_pointer)
>>> return -EINVAL;
>>> + if (amdgpu_kms_vram_lost(adev, fpriv))
>>> + return -ENODEV;
>>>
>>> switch (info->query) {
>>> case AMDGPU_INFO_VIRTUAL_RANGE: {
>>> @@ -779,6 +782,12 @@ void amdgpu_driver_lastclose_kms(struct
>>> drm_device *dev)
>>> vga_switcheroo_process_delayed_switch();
>>> }
>>>
>>> +bool amdgpu_kms_vram_lost(struct amdgpu_device *adev,
>>> + struct amdgpu_fpriv *fpriv)
>>> +{
>>> + return fpriv->vram_lost_counter != atomic_read(&adev-
>>>> vram_lost_counter);
>>> +}
>>> +
>>> /**
>>> * amdgpu_driver_open_kms - drm callback for open
>>> *
>>> @@ -833,6 +842,7 @@ int amdgpu_driver_open_kms(struct drm_device
>>> *dev, struct drm_file *file_priv)
>>>
>>> amdgpu_ctx_mgr_init(&fpriv->ctx_mgr);
>>>
>>> + fpriv->vram_lost_counter = atomic_read(&adev-
>>>> vram_lost_counter);
>>> file_priv->driver_priv = fpriv;
>>>
>>> out_suspend:
>>> --
>>> 1.9.1
>>>
>>> _______________________________________________
>>> amd-gfx mailing list
>>> amd-gfx@lists.freedesktop.org
>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>> _______________________________________________
>> amd-gfx mailing list
>> amd-gfx@lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>
>
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
next prev parent reply other threads:[~2017-05-24 2:20 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-05-16 9:25 [PATCH 1/4] drm/amdgpu: check if vram is lost v2 Chunming Zhou
[not found] ` <1494926750-1081-1-git-send-email-David1.Zhou-5C7GfCeVMHo@public.gmane.org>
2017-05-16 9:25 ` [PATCH 2/4] drm/amdgpu: return -ENODEV to user space when " Chunming Zhou
[not found] ` <1494926750-1081-2-git-send-email-David1.Zhou-5C7GfCeVMHo@public.gmane.org>
2017-05-23 15:08 ` Deucher, Alexander
[not found] ` <BN6PR12MB1652E1AFF691F58C92FC2CD7F7F90-/b2+HYfkarQqUD6E6FAiowdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org>
2017-05-23 15:16 ` Christian König
[not found] ` <69717c0b-b2c1-589a-c466-5d6be9518eda-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
2017-05-24 2:20 ` zhoucm1 [this message]
2017-05-16 9:25 ` [PATCH 3/4] drm/amdgpu: skip all jobs of guilty vm Chunming Zhou
2017-05-16 9:25 ` [PATCH 4/4] drm/amdgpu: reset fpriv vram_lost_counter Chunming Zhou
[not found] ` <1494926750-1081-4-git-send-email-David1.Zhou-5C7GfCeVMHo@public.gmane.org>
2017-05-17 1:18 ` Michel Dänzer
[not found] ` <58988726-543a-535a-3011-860d29b9f2da-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-17 3:04 ` zhoucm1
[not found] ` <591BBDA2.1070900-5C7GfCeVMHo@public.gmane.org>
2017-05-17 3:15 ` Michel Dänzer
[not found] ` <29fe2142-7fd1-e23a-49d9-c38dc685db92-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-17 4:28 ` zhoucm1
[not found] ` <591BD17C.8050903-5C7GfCeVMHo@public.gmane.org>
2017-05-17 6:57 ` Michel Dänzer
[not found] ` <7d87bc8e-9c09-ad25-de6e-dfbd8116bf6e-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-17 7:13 ` zhoucm1
[not found] ` <591BF825.6090505-5C7GfCeVMHo@public.gmane.org>
2017-05-17 8:01 ` Michel Dänzer
[not found] ` <31db7a30-dd98-5cb2-4125-187d3d0e2a49-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-17 8:40 ` Christian König
[not found] ` <7a302ebe-1de1-734f-fb21-aadcc7904d37-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
2017-05-17 8:46 ` zhoucm1
[not found] ` <591C0DFB.8030604-5C7GfCeVMHo@public.gmane.org>
2017-05-17 8:55 ` Michel Dänzer
2017-05-17 8:56 ` Christian König
[not found] ` <46582a1e-e019-34ac-1913-ed4a2a992e4c-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
2017-05-17 9:49 ` Marek Olšák
[not found] ` <CAAxE2A7sRZXx3MnRSO76DW=61X06nfVs2AHe_a-r+K+46tfJPw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-17 10:15 ` Zhou, David(ChunMing)
2017-05-16 10:49 ` [PATCH 1/4] drm/amdgpu: check if vram is lost v2 Christian König
[not found] ` <0c1d89c5-65ea-cdad-100a-80d0377b865c-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
2017-05-17 4:37 ` zhoucm1
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=5924EDFE.4010102@amd.com \
--to=david1.zhou-5c7gfcevmho@public.gmane.org \
--cc=Alexander.Deucher-5C7GfCeVMHo@public.gmane.org \
--cc=amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
--cc=deathsimple-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.