<html>
<head>
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p><br>
</p>
<br>
<div class="moz-cite-prefix">On 2018/9/15 17:23, Koenig, Christian
wrote:<br>
</div>
<blockquote type="cite"
cite="mid:68eb2303-ec42-470e-a060-487a14e2708c@email.android.com">
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
<meta content="text/html; charset=utf-8">
<div dir="auto">No, the problem is the function pointer analysis.
<div dir="auto"><br>
</div>
<div dir="auto">In other words the KIQ ring is sometimes used in
atomic and even interrupt context.</div>
<div dir="auto"><br>
</div>
<div dir="auto">But the UVD ring is never used in atomic
context.</div>
<div dir="auto"><br>
</div>
<div dir="auto">But I don't see a way a static analysis could
ever figure that out.</div>
<div dir="auto"><br>
</div>
</div>
</blockquote>
<br>
Okay, thanks for your explanation :)<br>
Besides, I find that amdgpu_virt_kiq_rreg() calls msleep(), so
mdelay() should be used instead.<br>
<br>
<br>
Best wishes,<br>
Jia-Ju Bai<br>
<br>
<blockquote type="cite"
cite="mid:68eb2303-ec42-470e-a060-487a14e2708c@email.android.com">
<div dir="auto"></div>
<div class="gmail_extra"><br>
<div class="gmail_quote">Am 15.09.2018 11:18 schrieb Jia-Ju Bai
<a class="moz-txt-link-rfc2396E" href="mailto:baijiaju1990@gmail.com"><baijiaju1990@gmail.com></a>:<br type="attribution">
</div>
</div>
<div>Sorry, I am still not clear why the call chain I proposed is
incorrect...<br>
<br>
I find a conditional in amdgpu_mm_wreg():<br>
<br>
if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
amdgpu_sriov_runtime(adev))<br>
return amdgpu_virt_kiq_wreg(adev, reg, v);<br>
<br>
Is amdgpu_virt_kiq_wreg() never called from WREG32() or
RREG32()?<br>
<br>
<br>
Best wishes,<br>
Jia-Ju Bai<br>
<br>
<br>
<div class="moz-cite-prefix">On 2018/9/15 17:10, Koenig,
Christian wrote:<br>
</div>
<blockquote type="cite">
<div dir="auto">
<div>amdgpu_ring_alloc() does call amdgpu_uvd_begin_use(),
but never in the call chain you proposed.</div>
<div dir="auto"><br>
</div>
<div dir="auto">Thinking about it I actually don't see a way
a statically analysis could ever figure that out.</div>
<div dir="auto"><br>
</div>
<div dir="auto">Christian.</div>
<div dir="auto">
<div class="gmail_extra" dir="auto"><br>
<div class="gmail_quote">Am 15.09.2018 11:05 schrieb
Jia-Ju Bai <a class="moz-txt-link-rfc2396E"
href="mailto:baijiaju1990@gmail.com"
moz-do-not-send="true">
<baijiaju1990@gmail.com></a>:<br
type="attribution">
<blockquote class="quote" style="margin:0 0 0 .8ex;
border-left:1px #ccc solid; padding-left:1ex">
<div>Thanks for your reply.<br>
<br>
<div>On 2018/9/15 17:01, Koenig, Christian wrote:<br>
</div>
<blockquote>
<div>
<div dir="auto">Sorry to say that but your
analysis tool is buggy.
<div dir="auto"><br>
</div>
<div dir="auto">The proposed call paths will
never trigger.</div>
</div>
</div>
</blockquote>
<br>
Could you please explain which piece of the call
path is incorrect?<br>
I am not very sure of my function pointer
analysis.<br>
Does amdgpu_ring_alloc() never calls
amdgpu_uvd_ring_begin_use()?<br>
<br>
Thanks in advance.<br>
<br>
<br>
Best wishes,<br>
Jia-Ju Bai<br>
<br>
<blockquote>
<div>
<div dir="auto">
<div dir="auto"><br>
</div>
<div dir="auto">Regards,</div>
<div dir="auto">Christian.</div>
</div>
<div><br>
<div>Am 15.09.2018 10:59 schrieb Jia-Ju Bai
<a href="mailto:baijiaju1990@gmail.com"
moz-do-not-send="true">
<baijiaju1990@gmail.com></a>:<br
type="attribution">
</div>
</div>
</div>
<font size="2"><span style="font-size:11pt">
<div>The driver may sleep with holding a
spinlock.<br>
The function call paths (from bottom to
top) in Linux-4.17 are:<br>
<br>
[FUNC] mutex_lock_nested<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c,
1477:<br>
mutex_lock_nested in
amdgpu_dpm_enable_uvd<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c,
1154:<br>
amdgpu_dpm_enable_uvd in
amdgpu_uvd_ring_begin_use<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c,
80:<br>
[FUNC_PTR]amdgpu_uvd_ring_begin_use
in amdgpu_ring_alloc<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c,
199:<br>
amdgpu_ring_alloc in
amdgpu_virt_kiq_wreg<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c, 207:<br>
amdgpu_virt_kiq_wreg in
amdgpu_mm_wreg<br>
drivers/gpu/drm/amd/amdgpu/vi.c, 207:<br>
amdgpu_mm_wreg in vi_gc_cac_rreg<br>
drivers/gpu/drm/amd/amdgpu/vi.c, 206:<br>
_raw_spin_lock_irqsave in
vi_gc_cac_rreg<br>
<br>
[FUNC] mutex_lock_nested<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c,
1477:<br>
mutex_lock_nested in
amdgpu_dpm_enable_uvd<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c,
1154:<br>
amdgpu_dpm_enable_uvd in
amdgpu_uvd_ring_begin_use<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c,
80:<br>
[FUNC_PTR]amdgpu_uvd_ring_begin_use
in amdgpu_ring_alloc<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c,
199:<br>
amdgpu_ring_alloc in
amdgpu_virt_kiq_wreg<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c, 207:<br>
amdgpu_virt_kiq_wreg in
amdgpu_mm_wreg<br>
drivers/gpu/drm/amd/amdgpu/soc15.c, 106:<br>
amdgpu_mm_wreg in soc15_pcie_rreg<br>
drivers/gpu/drm/amd/amdgpu/soc15.c, 105:<br>
_raw_spin_lock_irqsave in
soc15_pcie_rreg<br>
<br>
[FUNC] mutex_lock_nested<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c,
1477:<br>
mutex_lock_nested in
amdgpu_dpm_enable_uvd<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c,
1154:<br>
amdgpu_dpm_enable_uvd in
amdgpu_uvd_ring_begin_use<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c,
80:<br>
[FUNC_PTR]amdgpu_uvd_ring_begin_use
in amdgpu_ring_alloc<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c,
199:<br>
amdgpu_ring_alloc in
amdgpu_virt_kiq_wreg<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c, 207:<br>
amdgpu_virt_kiq_wreg in
amdgpu_mm_wreg<br>
drivers/gpu/drm/amd/amdgpu/cik.c, 139:<br>
amdgpu_mm_wreg in cik_uvd_ctx_wreg<br>
drivers/gpu/drm/amd/amdgpu/cik.c, 138:<br>
_raw_spin_lock_irqsave in
cik_uvd_ctx_wreg<br>
<br>
[FUNC] mutex_lock_nested<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c,
1477:<br>
mutex_lock_nested in
amdgpu_dpm_enable_uvd<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c,
1154:<br>
amdgpu_dpm_enable_uvd in
amdgpu_uvd_ring_begin_use<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c,
80:<br>
[FUNC_PTR]amdgpu_uvd_ring_begin_use
in amdgpu_ring_alloc<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c,
199:<br>
amdgpu_ring_alloc in
amdgpu_virt_kiq_wreg<br>
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c, 207:<br>
amdgpu_virt_kiq_wreg in
amdgpu_mm_wreg<br>
drivers/gpu/drm/amd/amdgpu/dce_v6_0.c,
126:<br>
amdgpu_mm_wreg in
dce_v6_0_audio_endpt_rreg<br>
drivers/gpu/drm/amd/amdgpu/dce_v6_0.c,
125:<br>
_raw_spin_lock_irqsave in
dce_v6_0_audio_endpt_rreg<br>
<br>
Note that [FUNC_PTR] means a function
pointer call is used.<br>
<br>
These bugs are found by my static analysis
tool DSAC.<br>
<br>
<br>
Best wishes,<br>
Jia-Ju Bai<br>
</div>
</span></font></blockquote>
<br>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</blockquote>
<br>
</div>
</blockquote>
<br>
</body>
</html>