[PATCH 0/2] virt: Convert to platform remove callback returning void

[PATCH 0/2] virt: Convert to platform remove callback returning void

[Uwe Kleine-König]

Hello,

this series converts the two platform drivers below drivers/virt that
make use of .remove() to use .remove_new() instead.

See commit 5c5a7680e67b ("platform: Provide a remove callback that
returns no value") for an extended explanation and the eventual goal.
The TL;DR; is to make it harder for driver authors to leak resources
without noticing. The drivers here get it right though and so can be
converted trivially.

This is merge window material. There doesn't seem to be a maintainer for
all of drivers/virt and I don't know how patch application works there
usally. The two patches are independent of each other so they can be
applied individually if necessary.

Best regards
Uwe

Uwe Kleine-König (2):
  virt: efi_secret: Convert to platform remove callback returning void
  virt: sev-guest: Convert to platform remove callback returning void

 drivers/virt/coco/efi_secret/efi_secret.c | 5 ++---
 drivers/virt/coco/sev-guest/sev-guest.c   | 6 ++----
 2 files changed, 4 insertions(+), 7 deletions(-)

base-commit: b85ea95d086471afb4ad062012a4d73cd328fa86
-- 
2.43.0

[PATCH 1/2] virt: efi_secret: Convert to platform remove callback returning void

[Uwe Kleine-König]

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is ignored (apart
from emitting a warning) and this typically results in resource leaks.

To improve here there is a quest to make the remove callback return
void. In the first step of this quest all drivers are converted to
.remove_new(), which already returns void. Eventually after all drivers
are converted, .remove_new() will be renamed to .remove().

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
---
 drivers/virt/coco/efi_secret/efi_secret.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/virt/coco/efi_secret/efi_secret.c b/drivers/virt/coco/efi_secret/efi_secret.c
index e700a5ef7043..cd29e66b1543 100644
--- a/drivers/virt/coco/efi_secret/efi_secret.c
+++ b/drivers/virt/coco/efi_secret/efi_secret.c
@@ -326,16 +326,15 @@ static int efi_secret_probe(struct platform_device *dev)
 	return ret;
 }
 
-static int efi_secret_remove(struct platform_device *dev)
+static void efi_secret_remove(struct platform_device *dev)
 {
 	efi_secret_securityfs_teardown(dev);
 	efi_secret_unmap_area();
-	return 0;
 }
 
 static struct platform_driver efi_secret_driver = {
 	.probe = efi_secret_probe,
-	.remove = efi_secret_remove,
+	.remove_new = efi_secret_remove,
 	.driver = {
 		.name = "efi_secret",
 	},
-- 
2.43.0

[PATCH 2/2] virt: sev-guest: Convert to platform remove callback returning void

[Uwe Kleine-König]

The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is ignored (apart
from emitting a warning) and this typically results in resource leaks.

To improve here there is a quest to make the remove callback return
void. In the first step of this quest all drivers are converted to
.remove_new(), which already returns void. Eventually after all drivers
are converted, .remove_new() will be renamed to .remove().

Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
---
 drivers/virt/coco/sev-guest/sev-guest.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/virt/coco/sev-guest/sev-guest.c b/drivers/virt/coco/sev-guest/sev-guest.c
index bc564adcf499..87f241825bc3 100644
--- a/drivers/virt/coco/sev-guest/sev-guest.c
+++ b/drivers/virt/coco/sev-guest/sev-guest.c
@@ -994,7 +994,7 @@ static int __init sev_guest_probe(struct platform_device *pdev)
 	return ret;
 }
 
-static int __exit sev_guest_remove(struct platform_device *pdev)
+static void __exit sev_guest_remove(struct platform_device *pdev)
 {
 	struct snp_guest_dev *snp_dev = platform_get_drvdata(pdev);
 
@@ -1003,8 +1003,6 @@ static int __exit sev_guest_remove(struct platform_device *pdev)
 	free_shared_pages(snp_dev->request, sizeof(struct snp_guest_msg));
 	deinit_crypto(snp_dev->crypto);
 	misc_deregister(&snp_dev->misc);
-
-	return 0;
 }
 
 /*
@@ -1013,7 +1011,7 @@ static int __exit sev_guest_remove(struct platform_device *pdev)
  * with the SEV-SNP support, it is named "sev-guest".
  */
 static struct platform_driver sev_guest_driver = {
-	.remove		= __exit_p(sev_guest_remove),
+	.remove_new	= __exit_p(sev_guest_remove),
 	.driver		= {
 		.name = "sev-guest",
 	},
-- 
2.43.0

Re: [PATCH 2/2] virt: sev-guest: Convert to platform remove callback returning void

[Tom Lendacky]

On 12/26/23 07:28, Uwe Kleine-König wrote:
> The .remove() callback for a platform driver returns an int which makes
> many driver authors wrongly assume it's possible to do error handling by
> returning an error code. However the value returned is ignored (apart
> from emitting a warning) and this typically results in resource leaks.
> 
> To improve here there is a quest to make the remove callback return
> void. In the first step of this quest all drivers are converted to
> .remove_new(), which already returns void. Eventually after all drivers
> are converted, .remove_new() will be renamed to .remove().
> 
> Trivially convert this driver from always returning zero in the remove
> callback to the void returning variant.
> 
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

Acked-by: Tom Lendacky <thomas.lendacky@amd.com>

> ---
>   drivers/virt/coco/sev-guest/sev-guest.c | 6 ++----
>   1 file changed, 2 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/virt/coco/sev-guest/sev-guest.c b/drivers/virt/coco/sev-guest/sev-guest.c
> index bc564adcf499..87f241825bc3 100644
> --- a/drivers/virt/coco/sev-guest/sev-guest.c
> +++ b/drivers/virt/coco/sev-guest/sev-guest.c
> @@ -994,7 +994,7 @@ static int __init sev_guest_probe(struct platform_device *pdev)
>   	return ret;
>   }
>   
> -static int __exit sev_guest_remove(struct platform_device *pdev)
> +static void __exit sev_guest_remove(struct platform_device *pdev)
>   {
>   	struct snp_guest_dev *snp_dev = platform_get_drvdata(pdev);
>   
> @@ -1003,8 +1003,6 @@ static int __exit sev_guest_remove(struct platform_device *pdev)
>   	free_shared_pages(snp_dev->request, sizeof(struct snp_guest_msg));
>   	deinit_crypto(snp_dev->crypto);
>   	misc_deregister(&snp_dev->misc);
> -
> -	return 0;
>   }
>   
>   /*
> @@ -1013,7 +1011,7 @@ static int __exit sev_guest_remove(struct platform_device *pdev)
>    * with the SEV-SNP support, it is named "sev-guest".
>    */
>   static struct platform_driver sev_guest_driver = {
> -	.remove		= __exit_p(sev_guest_remove),
> +	.remove_new	= __exit_p(sev_guest_remove),
>   	.driver		= {
>   		.name = "sev-guest",
>   	},

Re: [PATCH 1/2] virt: efi_secret: Convert to platform remove callback returning void

[Ard Biesheuvel]

On Tue, 26 Dec 2023 at 14:28, Uwe Kleine-König
<u.kleine-koenig@pengutronix.de> wrote:
>
> The .remove() callback for a platform driver returns an int which makes
> many driver authors wrongly assume it's possible to do error handling by
> returning an error code. However the value returned is ignored (apart
> from emitting a warning) and this typically results in resource leaks.
>
> To improve here there is a quest to make the remove callback return
> void. In the first step of this quest all drivers are converted to
> .remove_new(), which already returns void. Eventually after all drivers
> are converted, .remove_new() will be renamed to .remove().
>
> Trivially convert this driver from always returning zero in the remove
> callback to the void returning variant.
>
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

Acked-by: Ard Biesheuvel <ardb@kernel.org>

> ---
>  drivers/virt/coco/efi_secret/efi_secret.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/virt/coco/efi_secret/efi_secret.c b/drivers/virt/coco/efi_secret/efi_secret.c
> index e700a5ef7043..cd29e66b1543 100644
> --- a/drivers/virt/coco/efi_secret/efi_secret.c
> +++ b/drivers/virt/coco/efi_secret/efi_secret.c
> @@ -326,16 +326,15 @@ static int efi_secret_probe(struct platform_device *dev)
>         return ret;
>  }
>
> -static int efi_secret_remove(struct platform_device *dev)
> +static void efi_secret_remove(struct platform_device *dev)
>  {
>         efi_secret_securityfs_teardown(dev);
>         efi_secret_unmap_area();
> -       return 0;
>  }
>
>  static struct platform_driver efi_secret_driver = {
>         .probe = efi_secret_probe,
> -       .remove = efi_secret_remove,
> +       .remove_new = efi_secret_remove,
>         .driver = {
>                 .name = "efi_secret",
>         },
> --
> 2.43.0
>

Re: [PATCH 1/2] virt: efi_secret: Convert to platform remove callback returning void

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 1114 bytes --]

Hello,

On Tue, Dec 26, 2023 at 02:28:02PM +0100, Uwe Kleine-König wrote:
> The .remove() callback for a platform driver returns an int which makes
> many driver authors wrongly assume it's possible to do error handling by
> returning an error code. However the value returned is ignored (apart
> from emitting a warning) and this typically results in resource leaks.
> 
> To improve here there is a quest to make the remove callback return
> void. In the first step of this quest all drivers are converted to
> .remove_new(), which already returns void. Eventually after all drivers
> are converted, .remove_new() will be renamed to .remove().
> 
> Trivially convert this driver from always returning zero in the remove
> callback to the void returning variant.
> 
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

Is this patch still on someone's radar? If a resend would help, please
tell me.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH 1/2] virt: efi_secret: Convert to platform remove callback returning void

[Ard Biesheuvel]

On Wed, 6 Mar 2024 at 22:35, Uwe Kleine-König
<u.kleine-koenig@pengutronix.de> wrote:
>
> Hello,
>
> On Tue, Dec 26, 2023 at 02:28:02PM +0100, Uwe Kleine-König wrote:
> > The .remove() callback for a platform driver returns an int which makes
> > many driver authors wrongly assume it's possible to do error handling by
> > returning an error code. However the value returned is ignored (apart
> > from emitting a warning) and this typically results in resource leaks.
> >
> > To improve here there is a quest to make the remove callback return
> > void. In the first step of this quest all drivers are converted to
> > .remove_new(), which already returns void. Eventually after all drivers
> > are converted, .remove_new() will be renamed to .remove().
> >
> > Trivially convert this driver from always returning zero in the remove
> > callback to the void returning variant.
> >
> > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
>
> Is this patch still on someone's radar? If a resend would help, please
> tell me.
>

I've queued this up in the EFI tree.

Thanks,