<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style>
</head>
<body dir="ltr">
<p style="font-family:Arial;font-size:10pt;color:#0000FF;margin:5pt;" align="Left">
[AMD Official Use Only]<br>
</p>
<br>
<div>
<div style="font-family: Calibri, Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);">
<br>
</div>
<hr tabindex="-1" style="display:inline-block; width:98%">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" color="#000000" style="font-size:11pt"><b>From:</b> Koenig, Christian <Christian.Koenig@amd.com><br>
<b>Sent:</b> Wednesday, April 20, 2022 5:00 PM<br>
<b>To:</b> Wang, Yang(Kevin) <KevinYang.Wang@amd.com>; dri-devel@lists.freedesktop.org <dri-devel@lists.freedesktop.org>; amd-gfx@lists.freedesktop.org <amd-gfx@lists.freedesktop.org><br>
<b>Subject:</b> Re: [PATCH] drm/ttm: fix ttm tt init fail when size exceeds kmalloc limit</font>
<div> </div>
</div>
<div class="BodyFragment"><font size="2"><span style="font-size:11pt">
<div class="PlainText elementToProof">Am 20.04.22 um 10:56 schrieb Yang Wang:<br>
> if the __GFP_ZERO is set, the kvmalloc() can't fallback to use vmalloc()<br>
<br>
Hui what? Why should kvmalloc() not be able to fallback to vmalloc() <br>
when __GFP_ZERO is set?<br>
<br>
And even that is really the case then that sounds like a bug in kvmalloc().</div>
<div class="PlainText elementToProof"><br>
Regards,<br>
Christian.</div>
<div class="PlainText elementToProof"><br>
<div class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">[kevin]:</div>
</div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">it is really test case from libdrm amdgpu test, which try to allocate a big BO which will cause ttm tt init fail.</span><br>
</div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">it may be a kvmalloc() bug, and this patch can as a workaround in<span style="background-color:rgb(255, 255, 255);display:inline !important"> ttm</span> before
this issue fix.</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)"><br>
</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">void *kvmalloc_node(size_t size, gfp_t flags, int node) <br>
</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">{</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">...</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)"> if ((flags & GFP_KERNEL) != GFP_KERNEL)<br>
</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)"> return kmalloc_node(size, flags, node);<br>
</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">...</span></div>
<div class="PlainText elementToProof"><span class="PlainText" style="margin:0px;background-color:rgb(255, 255, 255)">}</span></div>
<div class="PlainText elementToProof"><br>
</div>
<div class="PlainText elementToProof">Best Regards,</div>
<div class="PlainText elementToProof">Kevin<br>
<br>
> to allocate memory, when request size is exceeds kmalloc limit, it will<br>
> cause allocate memory fail.<br>
><br>
> e.g: when ttm want to create a BO with 1TB size, it maybe fail.<br>
><br>
> Signed-off-by: Yang Wang <KevinYang.Wang@amd.com><br>
> ---<br>
> drivers/gpu/drm/ttm/ttm_tt.c | 14 +++++++++++---<br>
> 1 file changed, 11 insertions(+), 3 deletions(-)<br>
><br>
> diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c<br>
> index 79c870a3bef8..9f2f3e576b8d 100644<br>
> --- a/drivers/gpu/drm/ttm/ttm_tt.c<br>
> +++ b/drivers/gpu/drm/ttm/ttm_tt.c<br>
> @@ -97,9 +97,12 @@ int ttm_tt_create(struct ttm_buffer_object *bo, bool zero_alloc)<br>
> static int ttm_tt_alloc_page_directory(struct ttm_tt *ttm)<br>
> {<br>
> ttm->pages = kvmalloc_array(ttm->num_pages, sizeof(void*),<br>
> - GFP_KERNEL | __GFP_ZERO);<br>
> + GFP_KERNEL);<br>
> if (!ttm->pages)<br>
> return -ENOMEM;<br>
> +<br>
> + memset(ttm->pages, 0, ttm->num_pages * sizeof(void *));<br>
> +<br>
> return 0;<br>
> }<br>
> <br>
> @@ -108,10 +111,12 @@ static int ttm_dma_tt_alloc_page_directory(struct ttm_tt *ttm)<br>
> ttm->pages = kvmalloc_array(ttm->num_pages,<br>
> sizeof(*ttm->pages) +<br>
> sizeof(*ttm->dma_address),<br>
> - GFP_KERNEL | __GFP_ZERO);<br>
> + GFP_KERNEL);<br>
> if (!ttm->pages)<br>
> return -ENOMEM;<br>
> <br>
> + memset(ttm->pages, 0, ttm->num_pages * (sizeof(*ttm->pages) + sizeof(*ttm->dma_address)));<br>
> +<br>
> ttm->dma_address = (void *)(ttm->pages + ttm->num_pages);<br>
> return 0;<br>
> }<br>
> @@ -120,9 +125,12 @@ static int ttm_sg_tt_alloc_page_directory(struct ttm_tt *ttm)<br>
> {<br>
> ttm->dma_address = kvmalloc_array(ttm->num_pages,<br>
> sizeof(*ttm->dma_address),<br>
> - GFP_KERNEL | __GFP_ZERO);<br>
> + GFP_KERNEL);<br>
> if (!ttm->dma_address)<br>
> return -ENOMEM;<br>
> +<br>
> + memset(ttm->dma_address, 0, ttm->num_pages * sizeof(*ttm->dma_address));<br>
> +<br>
> return 0;<br>
> }<br>
> <br>
<br>
</div>
</span></font></div>
</div>
</body>
</html>