[PATCH] PCI: Remove the unused pci wrappers

[PATCH] PCI: Remove the unused pci wrappers

[Cai Huoqing]

The wrappers in include/linux/pci-dma-compat.h should go away,
so remove the unused pci wrappers.
Prefer using dma_xxx() instead of the pci wrappers pci_xxx().

Signed-off-by: Cai Huoqing <caihuoqing@baidu.com>
---
 include/linux/pci-dma-compat.h | 27 ---------------------------
 1 file changed, 27 deletions(-)

diff --git a/include/linux/pci-dma-compat.h b/include/linux/pci-dma-compat.h
index 249d4d7fbf18..33b316f38e1d 100644
--- a/include/linux/pci-dma-compat.h
+++ b/include/linux/pci-dma-compat.h
@@ -20,13 +20,6 @@ pci_alloc_consistent(struct pci_dev *hwdev, size_t size,
 	return dma_alloc_coherent(&hwdev->dev, size, dma_handle, GFP_ATOMIC);
 }
 
-static inline void *
-pci_zalloc_consistent(struct pci_dev *hwdev, size_t size,
-		      dma_addr_t *dma_handle)
-{
-	return dma_alloc_coherent(&hwdev->dev, size, dma_handle, GFP_ATOMIC);
-}
-
 static inline void
 pci_free_consistent(struct pci_dev *hwdev, size_t size,
 		    void *vaddr, dma_addr_t dma_handle)
@@ -89,26 +82,6 @@ pci_dma_sync_single_for_device(struct pci_dev *hwdev, dma_addr_t dma_handle,
 	dma_sync_single_for_device(&hwdev->dev, dma_handle, size, (enum dma_data_direction)direction);
 }
 
-static inline void
-pci_dma_sync_sg_for_cpu(struct pci_dev *hwdev, struct scatterlist *sg,
-		int nelems, int direction)
-{
-	dma_sync_sg_for_cpu(&hwdev->dev, sg, nelems, (enum dma_data_direction)direction);
-}
-
-static inline void
-pci_dma_sync_sg_for_device(struct pci_dev *hwdev, struct scatterlist *sg,
-		int nelems, int direction)
-{
-	dma_sync_sg_for_device(&hwdev->dev, sg, nelems, (enum dma_data_direction)direction);
-}
-
-static inline int
-pci_dma_mapping_error(struct pci_dev *pdev, dma_addr_t dma_addr)
-{
-	return dma_mapping_error(&pdev->dev, dma_addr);
-}
-
 #ifdef CONFIG_PCI
 static inline int pci_set_dma_mask(struct pci_dev *dev, u64 mask)
 {
-- 
2.25.1

Re: [PATCH] PCI: Remove the unused pci wrappers

[Christophe JAILLET]

Le 25/09/2021 à 15:52, Cai Huoqing a écrit :
> The wrappers in include/linux/pci-dma-compat.h should go away,
> so remove the unused pci wrappers.
> Prefer using dma_xxx() instead of the pci wrappers pci_xxx().
> 
> Signed-off-by: Cai Huoqing <caihuoqing@baidu.com>
> ---
>   include/linux/pci-dma-compat.h | 27 ---------------------------
>   1 file changed, 27 deletions(-)
> 
> diff --git a/include/linux/pci-dma-compat.h b/include/linux/pci-dma-compat.h
> index 249d4d7fbf18..33b316f38e1d 100644
> --- a/include/linux/pci-dma-compat.h
> +++ b/include/linux/pci-dma-compat.h
> @@ -20,13 +20,6 @@ pci_alloc_consistent(struct pci_dev *hwdev, size_t size,
>   	return dma_alloc_coherent(&hwdev->dev, size, dma_handle, GFP_ATOMIC);
>   }
>   
> -static inline void *
> -pci_zalloc_consistent(struct pci_dev *hwdev, size_t size,
> -		      dma_addr_t *dma_handle)
> -{
> -	return dma_alloc_coherent(&hwdev->dev, size, dma_handle, GFP_ATOMIC);
> -}
> -
>   static inline void
>   pci_free_consistent(struct pci_dev *hwdev, size_t size,
>   		    void *vaddr, dma_addr_t dma_handle)
> @@ -89,26 +82,6 @@ pci_dma_sync_single_for_device(struct pci_dev *hwdev, dma_addr_t dma_handle,
>   	dma_sync_single_for_device(&hwdev->dev, dma_handle, size, (enum dma_data_direction)direction);
>   }
>   
> -static inline void
> -pci_dma_sync_sg_for_cpu(struct pci_dev *hwdev, struct scatterlist *sg,
> -		int nelems, int direction)
> -{
> -	dma_sync_sg_for_cpu(&hwdev->dev, sg, nelems, (enum dma_data_direction)direction);
> -}
> -
> -static inline void
> -pci_dma_sync_sg_for_device(struct pci_dev *hwdev, struct scatterlist *sg,
> -		int nelems, int direction)
> -{
> -	dma_sync_sg_for_device(&hwdev->dev, sg, nelems, (enum dma_data_direction)direction);
> -}
> -
> -static inline int
> -pci_dma_mapping_error(struct pci_dev *pdev, dma_addr_t dma_addr)
> -{
> -	return dma_mapping_error(&pdev->dev, dma_addr);
> -}
> -
>   #ifdef CONFIG_PCI
>   static inline int pci_set_dma_mask(struct pci_dev *dev, u64 mask)
>   {
> 

Hi,

I'm not sure that this step is needed.
The whole pci-dma-compat.h is close to be obsolete. So axing just a part 
of it now is not that really useful.

After many patches, there is now just a few users of this deprecated 
API. All steps to finalize the job have already been posted [1] and a 
plan has been proposed by Arnd Bergmann to have the remaining ones 
merged in the tree [2].

Some of the message/fusion patches have been re-sent yesterday and today 
to the scsi maintainer in order to have them merged.

BTW, Cai Huoqing, thanks a lot for taking care and proposing the needed 
clean-ups in comments and log messages.
Looking at it was the next item of my to-do list, but I'm really pleased 
to get some help here to finalize the job once and for all.

CJ

[1]: 
https://lore.kernel.org/kernel-janitors/4cb6f731-7f4d-21d0-c88d-37664ea35002@wanadoo.fr/

[2]: 
https://lore.kernel.org/kernel-janitors/CAK8P3a2CBvw_GP372R+p8f4_pa82sMuQ5iHk4Nb2dJCzm_Fivw@mail.gmail.com/ 

Re: [PATCH] PCI: Remove the unused pci wrappers

[Cai Huoqing]

On 25 9月 21 17:14:05, Christophe JAILLET wrote:
> Le 25/09/2021 à 15:52, Cai Huoqing a écrit :
> > The wrappers in include/linux/pci-dma-compat.h should go away,
> > so remove the unused pci wrappers.
> > Prefer using dma_xxx() instead of the pci wrappers pci_xxx().
> > 
> > Signed-off-by: Cai Huoqing <caihuoqing@baidu.com>
> > ---
> >   include/linux/pci-dma-compat.h | 27 ---------------------------
> >   1 file changed, 27 deletions(-)
> > 
> > diff --git a/include/linux/pci-dma-compat.h b/include/linux/pci-dma-compat.h
> > index 249d4d7fbf18..33b316f38e1d 100644
> > --- a/include/linux/pci-dma-compat.h
> > +++ b/include/linux/pci-dma-compat.h
> > @@ -20,13 +20,6 @@ pci_alloc_consistent(struct pci_dev *hwdev, size_t size,
> >   	return dma_alloc_coherent(&hwdev->dev, size, dma_handle, GFP_ATOMIC);
> >   }
> > -static inline void *
> > -pci_zalloc_consistent(struct pci_dev *hwdev, size_t size,
> > -		      dma_addr_t *dma_handle)
> > -{
> > -	return dma_alloc_coherent(&hwdev->dev, size, dma_handle, GFP_ATOMIC);
> > -}
> > -
> >   static inline void
> >   pci_free_consistent(struct pci_dev *hwdev, size_t size,
> >   		    void *vaddr, dma_addr_t dma_handle)
> > @@ -89,26 +82,6 @@ pci_dma_sync_single_for_device(struct pci_dev *hwdev, dma_addr_t dma_handle,
> >   	dma_sync_single_for_device(&hwdev->dev, dma_handle, size, (enum dma_data_direction)direction);
> >   }
> > -static inline void
> > -pci_dma_sync_sg_for_cpu(struct pci_dev *hwdev, struct scatterlist *sg,
> > -		int nelems, int direction)
> > -{
> > -	dma_sync_sg_for_cpu(&hwdev->dev, sg, nelems, (enum dma_data_direction)direction);
> > -}
> > -
> > -static inline void
> > -pci_dma_sync_sg_for_device(struct pci_dev *hwdev, struct scatterlist *sg,
> > -		int nelems, int direction)
> > -{
> > -	dma_sync_sg_for_device(&hwdev->dev, sg, nelems, (enum dma_data_direction)direction);
> > -}
> > -
> > -static inline int
> > -pci_dma_mapping_error(struct pci_dev *pdev, dma_addr_t dma_addr)
> > -{
> > -	return dma_mapping_error(&pdev->dev, dma_addr);
> > -}
> > -
> >   #ifdef CONFIG_PCI
> >   static inline int pci_set_dma_mask(struct pci_dev *dev, u64 mask)
> >   {
> > 
> 
> Hi,
> 
> I'm not sure that this step is needed.
> The whole pci-dma-compat.h is close to be obsolete. So axing just a part of
> it now is not that really useful.
Thanks for your feedback.

If someone reuse these wrappers, that will not be checked by checkpatch.pl.
And I also will not add check condition to checkpatch.pl, because the wrappers
in include/linux/pci-dma-compat.h will go away soon.

Removing these API helps driver owner to avoid reusing these obsolete wrappers
without reviwing it manually, only works before pci-dma-compat.h is droped.

Thanks
Cai

> 
> After many patches, there is now just a few users of this deprecated API.
> All steps to finalize the job have already been posted [1] and a plan has
> been proposed by Arnd Bergmann to have the remaining ones merged in the tree
> [2]
it maybe cost serval weeks before all subtrees merge to linux-next.
> 
> Some of the message/fusion patches have been re-sent yesterday and today to
> the scsi maintainer in order to have them merged.
> 
> BTW, Cai Huoqing, thanks a lot for taking care and proposing the needed
> clean-ups in comments and log messages.
> Looking at it was the next item of my to-do list, but I'm really pleased to
> get some help here to finalize the job once and for all.
> 
> CJ
> 
> [1]: https://lore.kernel.org/kernel-janitors/4cb6f731-7f4d-21d0-c88d-37664ea35002@wanadoo.fr/
> 
> [2]: https://lore.kernel.org/kernel-janitors/CAK8P3a2CBvw_GP372R+p8f4_pa82sMuQ5iHk4Nb2dJCzm_Fivw@mail.gmail.com/
>