Re: [PATCH] media: mediatek: vcodec: using encoder device to alloc/free encoder memory

["Yunfei Dong (董云飞)" <Yunfei.Dong@mediatek.com>]

Hi AngeloGioacchino,

Thanks for your advice.
On Mon, 2023-10-09 at 14:15 +0200, AngeloGioacchino Del Regno wrote:
> Il 08/10/23 05:19, Yunfei Dong ha scritto:
> > Need to use encoder device to allocate/free encoder memory when
> > calling
> > mtk_vcodec_mem_alloc/mtk_vcodec_mem_free, or leading to below crash
> > log
> > when test encoder with decoder device.
> > 
> > pc : dma_alloc_attrs+0x44/0xf4
> > lr : mtk_vcodec_mem_alloc+0x50/0xa4 [mtk_vcodec_common]
> > sp : ffffffc0209f3990
> > x29: ffffffc0209f39a0 x28: ffffff8024102a18 x27: 0000000000000000
> > x26: 0000000000000000 x25: ffffffc00c06e2d8 x24: 0000000000000001
> > x23: 0000000000000cc0 x22: 0000000000000010 x21: 0000000000000800
> > x20: ffffff8024102a18 x19: 0000000000000000 x18: 0000000000000000
> > x17: 0000000000000009 x16: ffffffe389736a98 x15: 0000000000000078
> > x14: ffffffe389704434 x13: 0000000000000007 x12: ffffffe38a2b2560
> > x11: 0000000000000800 x10: 0000000000000004 x9 : ffffffe331f07484
> > x8 : 5400e9aef2395000 x7 : 0000000000000000 x6 : 000000000000003f
> > x5 : 0000000000000001 x4 : 0000000000000000 x3 : 0000000000000cc0
> > x2 : ffffff8024102a18 x1 : 0000000000000800 x0 : 0000000000000010
> > Call trace:
> >   dma_alloc_attrs+0x44/0xf4
> >   mtk_vcodec_mem_alloc+0x50/0xa4 [mtk_vcodec_common
> > 2819d3d601f3cd06c1f2213ac1b9995134441421]
> >   h264_enc_set_param+0x27c/0x378 [mtk_vcodec_enc
> > 772cc3d26c254e8cf54079451ef8d930d2eb4404]
> >   venc_if_set_param+0x4c/0x7c [mtk_vcodec_enc
> > 772cc3d26c254e8cf54079451ef8d930d2eb4404]
> >   vb2ops_venc_start_streaming+0x1bc/0x328 [mtk_vcodec_enc
> > 772cc3d26c254e8cf54079451ef8d930d2eb4404]
> >   vb2_start_streaming+0x64/0x12c
> >   vb2_core_streamon+0x114/0x158
> >   vb2_streamon+0x38/0x60
> >   v4l2_m2m_streamon+0x48/0x88
> >   v4l2_m2m_ioctl_streamon+0x20/0x2c
> >   v4l_streamon+0x2c/0x38
> >   __video_do_ioctl+0x2c4/0x3dc
> >   video_usercopy+0x404/0x934
> >   video_ioctl2+0x20/0x2c
> >   v4l2_ioctl+0x54/0x64
> >   v4l2_compat_ioctl32+0x90/0xa34
> >   __arm64_compat_sys_ioctl+0x128/0x13c
> >   invoke_syscall+0x4c/0x108
> >   el0_svc_common+0x98/0x104
> >   do_el0_svc_compat+0x28/0x34
> >   el0_svc_compat+0x2c/0x74
> >   el0t_32_sync_handler+0xa8/0xcc
> >   el0t_32_sync+0x194/0x198
> > Code: aa0003f6 aa0203f4 aa0103f5 f900
> > 
> > 'Fixes: 01abf5fbb081c ("media: mediatek: vcodec: separate struct
> > 'mtk_vcodec_ctx'")'
> > Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
> > ---
> >   .../mediatek/vcodec/common/mtk_vcodec_util.c  | 66
> > ++++++++++++++++++-
> >   1 file changed, 64 insertions(+), 2 deletions(-)
> > 
> > diff --git
> > a/drivers/media/platform/mediatek/vcodec/common/mtk_vcodec_util.c
> > b/drivers/media/platform/mediatek/vcodec/common/mtk_vcodec_util.c
> > index 908602031fd0..62bb7290c56d 100644
> > ---
> > a/drivers/media/platform/mediatek/vcodec/common/mtk_vcodec_util.c
> > +++
> > b/drivers/media/platform/mediatek/vcodec/common/mtk_vcodec_util.c
> > @@ -45,7 +45,7 @@ int mtk_vcodec_write_vdecsys(struct
> > mtk_vcodec_dec_ctx *ctx, unsigned int reg,
> >   }
> >   EXPORT_SYMBOL(mtk_vcodec_write_vdecsys);
> >   
> > -int mtk_vcodec_mem_alloc(void *priv, struct mtk_vcodec_mem *mem)
> > +static int mtk_vcodec_mem_dec_alloc(void *priv, struct
> > mtk_vcodec_mem *mem)
> >   {
> >   	unsigned long size = mem->size;
> >   	struct mtk_vcodec_dec_ctx *ctx = priv;
> > @@ -64,9 +64,39 @@ int mtk_vcodec_mem_alloc(void *priv, struct
> > mtk_vcodec_mem *mem)
> >   
> >   	return 0;
> >   }
> > +
> > +static int mtk_vcodec_mem_enc_alloc(void *priv, struct
> > mtk_vcodec_mem *mem)
> > +{
> > +	unsigned long size = mem->size;
> > +	struct mtk_vcodec_enc_ctx *ctx = priv;
> > +	struct device *dev = &ctx->dev->plat_dev->dev;
> > +
> > +	mem->va = dma_alloc_coherent(dev, size, &mem->dma_addr,
> > GFP_KERNEL);
> > +	if (!mem->va) {
> > +		mtk_v4l2_venc_err(ctx, "%s dma_alloc size=%ld failed!",
> > dev_name(dev), size);
> > +		return -ENOMEM;
> > +	}
> > +
> > +	mtk_v4l2_venc_dbg(3, ctx, "[%d]  - va      = %p", ctx->id, mem-
> > >va);
> > +	mtk_v4l2_venc_dbg(3, ctx, "[%d]  - dma     = 0x%lx", ctx->id,
> > +			  (unsigned long)mem->dma_addr);
> > +	mtk_v4l2_venc_dbg(3, ctx, "[%d]    size = 0x%lx", ctx->id,
> > size);
> > +
> > +	return 0;
> > +}
> > +
> > +int mtk_vcodec_mem_alloc(void *priv, struct mtk_vcodec_mem *mem)
> > +{
> > +	enum mtk_instance_type inst_type = *((unsigned int *)priv);
> 
> I agree in that the first member of both mtk_vcodec_{enc,dec}_ctx is
> this
> enum mtk_instance_type, but no, you're not passing that as priv:
> you're actually
> passing one of the two structures, and this would add up complexity
> in human
> readability, and would also open the road for mistakes.
> 
> You should at this point pass the instance type to the function.
> 
> int mtk_vcodec_mem_alloc(void *priv, enum mtk_instance_type /* or
> int, but I prefer 
> enum */ inst_type, struct mtk_vcodec_mem *mem)
> 
> ...and then, since the only difference between the two `alloc`
> function is just
> only about debugging logs, you could commonize the allocation part...
> an idea:
> 
> struct platform_device *pdev;
> unsigned long size;
> int id;
> 
> 
Will rewrite allocate and free memory interface in patch v2.

Best Regards,
Yunfei Dong
> if (inst_type == MTK_INST_DECODER) {
> 	struct mtk_vcodec_enc_ctx *enc_ctx = priv;
> 	pdev = enc_ctx->dev->plat_dev;
> 	id = enc_ctx->id;
> } else {
> 	struct mtk_vcodec_dec_ctx *dec_ctx = priv;
> 	pdev = dec_ctx->dev->plat_dev;
> 	id = dec_ctx->id;
> }
> 
> mem->va = dma_alloc_coherent(&pdev->dev, etc.....)
> if (!mem->va) {
> 	mtk_v4l2_err(pdev, .....);
> }
> 
> mtk_v4l2_debug(&pdev->dev, 3, "[%d]  - va      = %p", id, mem->va);
> 
> ...you wouldn't even need one function for enc_alloc and one for
> dec_alloc, as
> the logic would be pretty much commonized like this.
> 
> > +
> > +	if (inst_type == MTK_INST_ENCODER)
> > +		return mtk_vcodec_mem_enc_alloc(priv, mem);
> > +	else
> > +		return mtk_vcodec_mem_dec_alloc(priv, mem);
> > +}
> >   EXPORT_SYMBOL(mtk_vcodec_mem_alloc);
> >   
> > -void mtk_vcodec_mem_free(void *priv, struct mtk_vcodec_mem *mem)
> > +static void mtk_vcodec_mem_dec_free(void *priv, struct
> > mtk_vcodec_mem *mem)
> >   {
> >   	unsigned long size = mem->size;
> >   	struct mtk_vcodec_dec_ctx *ctx = priv;
> > @@ -87,6 +117,38 @@ void mtk_vcodec_mem_free(void *priv, struct
> > mtk_vcodec_mem *mem)
> >   	mem->dma_addr = 0;
> >   	mem->size = 0;
> >   }
> > +
> > +static void mtk_vcodec_mem_enc_free(void *priv, struct
> > mtk_vcodec_mem *mem)
> > +{
> > +	unsigned long size = mem->size;
> > +	struct mtk_vcodec_enc_ctx *ctx = priv;
> > +	struct device *dev = &ctx->dev->plat_dev->dev;
> > +
> > +	if (!mem->va) {
> > +		mtk_v4l2_venc_err(ctx, "%s dma_free size=%ld failed!",
> > dev_name(dev), size);
> > +		return;
> > +	}
> > +
> > +	mtk_v4l2_venc_dbg(3, ctx, "[%d]  - va      = %p", ctx->id, mem-
> > >va);
> > +	mtk_v4l2_venc_dbg(3, ctx, "[%d]  - dma     = 0x%lx", ctx->id,
> > +			  (unsigned long)mem->dma_addr);
> > +	mtk_v4l2_venc_dbg(3, ctx, "[%d]    size = 0x%lx", ctx->id,
> > size);
> > +
> > +	dma_free_coherent(dev, size, mem->va, mem->dma_addr);
> > +	mem->va = NULL;
> > +	mem->dma_addr = 0;
> > +	mem->size = 0;
> > +}
> > +
> > +void mtk_vcodec_mem_free(void *priv, struct mtk_vcodec_mem *mem)
> > +{
> > +	enum mtk_instance_type inst_type = *((unsigned int *)priv);
> > +
> > +	if (inst_type == MTK_INST_ENCODER)
> > +		mtk_vcodec_mem_enc_free(priv, mem);
> > +	else
> > +		mtk_vcodec_mem_dec_free(priv, mem);
> 
> same comments apply to the xxxx_free() function as well.
> You're in the "common" scope here, so you wouldn't be forced to use
> the enc/dec
> specific logging functions, as you can use the more generic
> mtk_v4l2_{debug,err}()
> and mtk_vcodec_{debug,err}() functions here, which are *perfect* for
> the common
> code.
> 
> Cheers,
> Angelo
> 
>