[PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[Souptick Joarder]

Use new return type vm_fault_t for fault handler. For
now, this is just documenting that the function returns
a VM_FAULT value rather than an errno. Once all instances
are converted, vm_fault_t will become a distinct type.

Reference id -> 1c8f422059ae ("mm: change return type to
vm_fault_t")

Previously vm_insert_page() returns err which driver
mapped into VM_FAULT_* type. The new function 
vmf_insert_page() will replace this inefficiency by
returning VM_FAULT_* type.

Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
---
 drivers/gpu/drm/tegra/gem.c | 18 ++----------------
 1 file changed, 2 insertions(+), 16 deletions(-)

diff --git a/drivers/gpu/drm/tegra/gem.c b/drivers/gpu/drm/tegra/gem.c
index 49b9bf2..6121493 100644
--- a/drivers/gpu/drm/tegra/gem.c
+++ b/drivers/gpu/drm/tegra/gem.c
@@ -422,14 +422,13 @@ int tegra_bo_dumb_create(struct drm_file *file, struct drm_device *drm,
 	return 0;
 }

-static int tegra_bo_fault(struct vm_fault *vmf)
+static vm_fault_t tegra_bo_fault(struct vm_fault *vmf)
 {
 	struct vm_area_struct *vma = vmf->vma;
 	struct drm_gem_object *gem = vma->vm_private_data;
 	struct tegra_bo *bo = to_tegra_bo(gem);
 	struct page *page;
 	pgoff_t offset;
-	int err;

 	if (!bo->pages)
 		return VM_FAULT_SIGBUS;
@@ -437,20 +436,7 @@ static int tegra_bo_fault(struct vm_fault *vmf)
 	offset = (vmf->address - vma->vm_start) >> PAGE_SHIFT;
 	page = bo->pages[offset];

-	err = vm_insert_page(vma, vmf->address, page);
-	switch (err) {
-	case -EAGAIN:
-	case 0:
-	case -ERESTARTSYS:
-	case -EINTR:
-	case -EBUSY:
-		return VM_FAULT_NOPAGE;
-
-	case -ENOMEM:
-		return VM_FAULT_OOM;
-	}
-
-	return VM_FAULT_SIGBUS;
+	return vmf_insert_page(vma, vmf->address, page);
 }

 const struct vm_operations_struct tegra_bo_vm_ops = {
--
1.9.1

Re: [PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 2147 bytes --]

On Tue, Apr 17, 2018 at 07:17:55PM +0530, Souptick Joarder wrote:
> Use new return type vm_fault_t for fault handler. For
> now, this is just documenting that the function returns
> a VM_FAULT value rather than an errno. Once all instances
> are converted, vm_fault_t will become a distinct type.
> 
> Reference id -> 1c8f422059ae ("mm: change return type to
> vm_fault_t")
> 
> Previously vm_insert_page() returns err which driver
> mapped into VM_FAULT_* type. The new function 
> vmf_insert_page() will replace this inefficiency by
> returning VM_FAULT_* type.
> 
> Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
> ---
>  drivers/gpu/drm/tegra/gem.c | 18 ++----------------
>  1 file changed, 2 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/gpu/drm/tegra/gem.c b/drivers/gpu/drm/tegra/gem.c
> index 49b9bf2..6121493 100644
> --- a/drivers/gpu/drm/tegra/gem.c
> +++ b/drivers/gpu/drm/tegra/gem.c
> @@ -422,14 +422,13 @@ int tegra_bo_dumb_create(struct drm_file *file, struct drm_device *drm,
>  	return 0;
>  }
> 
> -static int tegra_bo_fault(struct vm_fault *vmf)
> +static vm_fault_t tegra_bo_fault(struct vm_fault *vmf)
>  {
>  	struct vm_area_struct *vma = vmf->vma;
>  	struct drm_gem_object *gem = vma->vm_private_data;
>  	struct tegra_bo *bo = to_tegra_bo(gem);
>  	struct page *page;
>  	pgoff_t offset;
> -	int err;
> 
>  	if (!bo->pages)
>  		return VM_FAULT_SIGBUS;
> @@ -437,20 +436,7 @@ static int tegra_bo_fault(struct vm_fault *vmf)
>  	offset = (vmf->address - vma->vm_start) >> PAGE_SHIFT;
>  	page = bo->pages[offset];
> 
> -	err = vm_insert_page(vma, vmf->address, page);
> -	switch (err) {
> -	case -EAGAIN:
> -	case 0:
> -	case -ERESTARTSYS:
> -	case -EINTR:
> -	case -EBUSY:
> -		return VM_FAULT_NOPAGE;
> -
> -	case -ENOMEM:
> -		return VM_FAULT_OOM;
> -	}
> -
> -	return VM_FAULT_SIGBUS;
> +	return vmf_insert_page(vma, vmf->address, page);
>  }

This new function returns VM_FAULT_NOPAGE only for 0 and -EBUSY, whereas
we used to return VM_FAULT_NOPAGE for -EAGAIN, -ERESTARTSYS and -EINTR
as well. Was this previously wrong?

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[Souptick Joarder]

> This new function returns VM_FAULT_NOPAGE only for 0 and -EBUSY, whereas
> we used to return VM_FAULT_NOPAGE for -EAGAIN, -ERESTARTSYS and -EINTR
> as well.

Previously vm_insert_page unable to return VM_FAULT_ type due to which
different drivers
have their own mapping from err to VM_FAULT_ type. With new
vmf_insert_page() introduce
in 4.17-rc1 we have a common mapping for err to VM_FAULT_ which remain
same for every
drivers/file systems.

Was this previously wrong?

I think Matthew is the right person to answer this.

Re: [PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[Matthew Wilcox]

On Wed, Apr 18, 2018 at 10:24:39AM +0200, Thierry Reding wrote:
> > @@ -437,20 +436,7 @@ static int tegra_bo_fault(struct vm_fault *vmf)
> >  	offset = (vmf->address - vma->vm_start) >> PAGE_SHIFT;
> >  	page = bo->pages[offset];
> > 
> > -	err = vm_insert_page(vma, vmf->address, page);
> > -	switch (err) {
> > -	case -EAGAIN:
> > -	case 0:
> > -	case -ERESTARTSYS:
> > -	case -EINTR:
> > -	case -EBUSY:
> > -		return VM_FAULT_NOPAGE;
> > -
> > -	case -ENOMEM:
> > -		return VM_FAULT_OOM;
> > -	}
> > -
> > -	return VM_FAULT_SIGBUS;
> > +	return vmf_insert_page(vma, vmf->address, page);
> >  }
> 
> This new function returns VM_FAULT_NOPAGE only for 0 and -EBUSY, whereas
> we used to return VM_FAULT_NOPAGE for -EAGAIN, -ERESTARTSYS and -EINTR
> as well. Was this previously wrong?

Not so much wrong as unnecessary.  vm_insert_page() can't return -EAGAIN,
-ERESTARTSYS or -EINTR.

Re: [PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[Souptick Joarder]

On Mon, Apr 23, 2018 at 4:24 PM, Matthew Wilcox <willy@infradead.org> wrote:
> On Wed, Apr 18, 2018 at 10:24:39AM +0200, Thierry Reding wrote:
>> > @@ -437,20 +436,7 @@ static int tegra_bo_fault(struct vm_fault *vmf)
>> >     offset = (vmf->address - vma->vm_start) >> PAGE_SHIFT;
>> >     page = bo->pages[offset];
>> >
>> > -   err = vm_insert_page(vma, vmf->address, page);
>> > -   switch (err) {
>> > -   case -EAGAIN:
>> > -   case 0:
>> > -   case -ERESTARTSYS:
>> > -   case -EINTR:
>> > -   case -EBUSY:
>> > -           return VM_FAULT_NOPAGE;
>> > -
>> > -   case -ENOMEM:
>> > -           return VM_FAULT_OOM;
>> > -   }
>> > -
>> > -   return VM_FAULT_SIGBUS;
>> > +   return vmf_insert_page(vma, vmf->address, page);
>> >  }
>>
>> This new function returns VM_FAULT_NOPAGE only for 0 and -EBUSY, whereas
>> we used to return VM_FAULT_NOPAGE for -EAGAIN, -ERESTARTSYS and -EINTR
>> as well. Was this previously wrong?
>
> Not so much wrong as unnecessary.  vm_insert_page() can't return -EAGAIN,
> -ERESTARTSYS or -EINTR.

If no further comment on this patch, We would like to get this patch
queued for 4.18.

Re: [PATCH] gpu: drm: tegra: Adding new typedef vm_fault_t

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 786 bytes --]

On Tue, Apr 17, 2018 at 07:17:55PM +0530, Souptick Joarder wrote:
> Use new return type vm_fault_t for fault handler. For
> now, this is just documenting that the function returns
> a VM_FAULT value rather than an errno. Once all instances
> are converted, vm_fault_t will become a distinct type.
> 
> Reference id -> 1c8f422059ae ("mm: change return type to
> vm_fault_t")
> 
> Previously vm_insert_page() returns err which driver
> mapped into VM_FAULT_* type. The new function 
> vmf_insert_page() will replace this inefficiency by
> returning VM_FAULT_* type.
> 
> Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
> ---
>  drivers/gpu/drm/tegra/gem.c | 18 ++----------------
>  1 file changed, 2 insertions(+), 16 deletions(-)

Applied, thanks.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]