[PATCH 00/15] PCI: Convert to platform remove callback returning void

[PATCH 00/15] PCI: Convert to platform remove callback returning void

[Uwe Kleine-König]

Hello,

this series adapts the platform drivers below drivers/pci to use the
.remove_new() callback. Compared to the traditional .remove() callback
.remove_new() returns no value. This is a good thing because the driver core
doesn't (and cannot) cope for errors during remove. The only effect of a
non-zero return value in .remove() is that the driver core emits a warning. The
device is removed anyhow and an early return from .remove() usually yields a
resource leak.

By changing the remove callback to return void driver authors cannot
reasonably assume any more that there is some kind of cleanup later.

All drivers were easy to convert as they all returned zero in their
remove callback. Only for iproc the conversion wasn't trivial, the other
were converted using coccinelle.

There are no interdependencies between these patches. So even if there
are some concerns for individual patches, I ask you to apply the
remaining set. Then I only have to care for the review feedback of the
refused patches. (Having said that I don't expect any serious objection,
just things like squashing or separating patches, or maybe I picked a
wrong subject prefix.)

Best regards
Uwe

Uwe Kleine-König (15):
  PCI: aardvark: Convert to platform remove callback returning void
  PCI: altera: Convert to platform remove callback returning void
  PCI: altera-msi: Convert to platform remove callback returning void
  PCI: brcmstb: Convert to platform remove callback returning void
  PCI: cadence/j721e: Convert to platform remove callback returning void
  PCI: dwc/*: Convert to platform remove callback returning void
  PCI: hisi-error: Convert to platform remove callback returning void
  PCI: iproc: Convert to platform remove callback returning void
  PCI: mediatek: Convert to platform remove callback returning void
  PCI: mediatek-gen3: Convert to platform remove callback returning void
  PCI: mt7621: Convert to platform remove callback returning void
  PCI: mvebu: Convert to platform remove callback returning void
  PCI: rockchip-host: Convert to platform remove callback returning void
  PCI: tegra: Convert to platform remove callback returning void
  PCI: xgene-msi: Convert to platform remove callback returning void

 drivers/pci/controller/cadence/pci-j721e.c   | 6 ++----
 drivers/pci/controller/dwc/pcie-bt1.c        | 6 ++----
 drivers/pci/controller/dwc/pcie-histb.c      | 6 ++----
 drivers/pci/controller/dwc/pcie-intel-gw.c   | 6 ++----
 drivers/pci/controller/dwc/pcie-qcom-ep.c    | 8 +++-----
 drivers/pci/controller/dwc/pcie-tegra194.c   | 8 +++-----
 drivers/pci/controller/pci-aardvark.c        | 6 ++----
 drivers/pci/controller/pci-mvebu.c           | 6 ++----
 drivers/pci/controller/pci-tegra.c           | 6 ++----
 drivers/pci/controller/pci-xgene-msi.c       | 6 ++----
 drivers/pci/controller/pcie-altera-msi.c     | 5 ++---
 drivers/pci/controller/pcie-altera.c         | 6 ++----
 drivers/pci/controller/pcie-brcmstb.c        | 6 ++----
 drivers/pci/controller/pcie-hisi-error.c     | 6 ++----
 drivers/pci/controller/pcie-iproc-platform.c | 6 +++---
 drivers/pci/controller/pcie-iproc.c          | 4 +---
 drivers/pci/controller/pcie-iproc.h          | 2 +-
 drivers/pci/controller/pcie-mediatek-gen3.c  | 6 ++----
 drivers/pci/controller/pcie-mediatek.c       | 6 ++----
 drivers/pci/controller/pcie-mt7621.c         | 6 ++----
 drivers/pci/controller/pcie-rockchip-host.c  | 6 ++----
 21 files changed, 43 insertions(+), 80 deletions(-)

base-commit: fe15c26ee26efa11741a7b632e9f23b01aca4cc6
-- 
2.39.2

[PATCH 05/15] PCI: cadence/j721e: 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 (mostly) ignored
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.

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/pci/controller/cadence/pci-j721e.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/pci/controller/cadence/pci-j721e.c b/drivers/pci/controller/cadence/pci-j721e.c
index cc83a8925ce0..e70213c9060a 100644
--- a/drivers/pci/controller/cadence/pci-j721e.c
+++ b/drivers/pci/controller/cadence/pci-j721e.c
@@ -542,7 +542,7 @@ static int j721e_pcie_probe(struct platform_device *pdev)
 	return ret;
 }
 
-static int j721e_pcie_remove(struct platform_device *pdev)
+static void j721e_pcie_remove(struct platform_device *pdev)
 {
 	struct j721e_pcie *pcie = platform_get_drvdata(pdev);
 	struct cdns_pcie *cdns_pcie = pcie->cdns_pcie;
@@ -552,13 +552,11 @@ static int j721e_pcie_remove(struct platform_device *pdev)
 	cdns_pcie_disable_phy(cdns_pcie);
 	pm_runtime_put(dev);
 	pm_runtime_disable(dev);
-
-	return 0;
 }
 
 static struct platform_driver j721e_pcie_driver = {
 	.probe  = j721e_pcie_probe,
-	.remove = j721e_pcie_remove,
+	.remove_new = j721e_pcie_remove,
 	.driver = {
 		.name	= "j721e-pcie",
 		.of_match_table = of_j721e_pcie_match,
-- 
2.39.2

Re: [PATCH 00/15] PCI: Convert to platform remove callback returning void

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 1699 bytes --]

Hello Bjorn,

On Tue, Mar 21, 2023 at 08:31:53PM +0100, Uwe Kleine-König wrote:
> this series adapts the platform drivers below drivers/pci to use the
> .remove_new() callback. Compared to the traditional .remove() callback
> .remove_new() returns no value. This is a good thing because the driver core
> doesn't (and cannot) cope for errors during remove. The only effect of a
> non-zero return value in .remove() is that the driver core emits a warning. The
> device is removed anyhow and an early return from .remove() usually yields a
> resource leak.
> 
> By changing the remove callback to return void driver authors cannot
> reasonably assume any more that there is some kind of cleanup later.
> 
> All drivers were easy to convert as they all returned zero in their
> remove callback. Only for iproc the conversion wasn't trivial, the other
> were converted using coccinelle.
> 
> There are no interdependencies between these patches. So even if there
> are some concerns for individual patches, I ask you to apply the
> remaining set. Then I only have to care for the review feedback of the
> refused patches. (Having said that I don't expect any serious objection,
> just things like squashing or separating patches, or maybe I picked a
> wrong subject prefix.)

These patches wait for application for quite some time now. They apply
just fine to v6.4-rc1 and next/master. Would be great to get them in
during the next merge window and ideally give them some time in next
before.

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 00/15] PCI: Convert to platform remove callback returning void

[Bjorn Helgaas]

On Tue, May 30, 2023 at 04:07:42PM +0200, Uwe Kleine-König wrote:
> Hello Bjorn,
> 
> On Tue, Mar 21, 2023 at 08:31:53PM +0100, Uwe Kleine-König wrote:
> > this series adapts the platform drivers below drivers/pci to use the
> > .remove_new() callback. Compared to the traditional .remove() callback
> > .remove_new() returns no value. This is a good thing because the driver core
> > doesn't (and cannot) cope for errors during remove. The only effect of a
> > non-zero return value in .remove() is that the driver core emits a warning. The
> > device is removed anyhow and an early return from .remove() usually yields a
> > resource leak.
> > 
> > By changing the remove callback to return void driver authors cannot
> > reasonably assume any more that there is some kind of cleanup later.
> > 
> > All drivers were easy to convert as they all returned zero in their
> > remove callback. Only for iproc the conversion wasn't trivial, the other
> > were converted using coccinelle.
> > 
> > There are no interdependencies between these patches. So even if there
> > are some concerns for individual patches, I ask you to apply the
> > remaining set. Then I only have to care for the review feedback of the
> > refused patches. (Having said that I don't expect any serious objection,
> > just things like squashing or separating patches, or maybe I picked a
> > wrong subject prefix.)
> 
> These patches wait for application for quite some time now. They apply
> just fine to v6.4-rc1 and next/master. Would be great to get them in
> during the next merge window and ideally give them some time in next
> before.

Thanks, these seem fine to me, and Lorenzo normally takes care of
drivers/pci/controller/.  Lorenzo, if it's easier to have me apply
them, that's fine, too, just let me know.

The only tweaks I would make would be:

  PCI: j721e: Convert to platform remove callback returning void
  PCI: dwc: Convert to platform remove callback returning void

to match the git history.

Re: [PATCH 00/15] PCI: Convert to platform remove callback returning void

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 2641 bytes --]

Hello,

[dropped Tom Joseph and Joyce Ooi from Cc:, my last mail in this thread
resulted in a bounce for their addresses.]

On Fri, Jun 02, 2023 at 04:37:34PM -0500, Bjorn Helgaas wrote:
> On Tue, May 30, 2023 at 04:07:42PM +0200, Uwe Kleine-König wrote:
> > Hello Bjorn,
> > 
> > On Tue, Mar 21, 2023 at 08:31:53PM +0100, Uwe Kleine-König wrote:
> > > this series adapts the platform drivers below drivers/pci to use the
> > > .remove_new() callback. Compared to the traditional .remove() callback
> > > .remove_new() returns no value. This is a good thing because the driver core
> > > doesn't (and cannot) cope for errors during remove. The only effect of a
> > > non-zero return value in .remove() is that the driver core emits a warning. The
> > > device is removed anyhow and an early return from .remove() usually yields a
> > > resource leak.
> > > 
> > > By changing the remove callback to return void driver authors cannot
> > > reasonably assume any more that there is some kind of cleanup later.
> > > 
> > > All drivers were easy to convert as they all returned zero in their
> > > remove callback. Only for iproc the conversion wasn't trivial, the other
> > > were converted using coccinelle.
> > > 
> > > There are no interdependencies between these patches. So even if there
> > > are some concerns for individual patches, I ask you to apply the
> > > remaining set. Then I only have to care for the review feedback of the
> > > refused patches. (Having said that I don't expect any serious objection,
> > > just things like squashing or separating patches, or maybe I picked a
> > > wrong subject prefix.)
> > 
> > These patches wait for application for quite some time now. They apply
> > just fine to v6.4-rc1 and next/master. Would be great to get them in
> > during the next merge window and ideally give them some time in next
> > before.
> 
> Thanks, these seem fine to me, and Lorenzo normally takes care of
> drivers/pci/controller/.  Lorenzo, if it's easier to have me apply
> them, that's fine, too, just let me know.
> 
> The only tweaks I would make would be:
> 
>   PCI: j721e: Convert to platform remove callback returning void
>   PCI: dwc: Convert to platform remove callback returning void

If it's easier for you (or Lorenzo) I can resend with these tweaks.
Otherwise if these are adapted when applying them, that's fine for me,
too. Just tell me if I should do anything here.

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 00/15] PCI: Convert to platform remove callback returning void

[Pali Rohár]

On Friday 02 June 2023 16:37:34 Bjorn Helgaas wrote:
> On Tue, May 30, 2023 at 04:07:42PM +0200, Uwe Kleine-König wrote:
> > Hello Bjorn,
> > 
> > On Tue, Mar 21, 2023 at 08:31:53PM +0100, Uwe Kleine-König wrote:
> > > this series adapts the platform drivers below drivers/pci to use the
> > > .remove_new() callback. Compared to the traditional .remove() callback
> > > .remove_new() returns no value. This is a good thing because the driver core
> > > doesn't (and cannot) cope for errors during remove. The only effect of a
> > > non-zero return value in .remove() is that the driver core emits a warning. The
> > > device is removed anyhow and an early return from .remove() usually yields a
> > > resource leak.
> > > 
> > > By changing the remove callback to return void driver authors cannot
> > > reasonably assume any more that there is some kind of cleanup later.
> > > 
> > > All drivers were easy to convert as they all returned zero in their
> > > remove callback. Only for iproc the conversion wasn't trivial, the other
> > > were converted using coccinelle.
> > > 
> > > There are no interdependencies between these patches. So even if there
> > > are some concerns for individual patches, I ask you to apply the
> > > remaining set. Then I only have to care for the review feedback of the
> > > refused patches. (Having said that I don't expect any serious objection,
> > > just things like squashing or separating patches, or maybe I picked a
> > > wrong subject prefix.)
> > 
> > These patches wait for application for quite some time now. They apply
> > just fine to v6.4-rc1 and next/master. Would be great to get them in
> > during the next merge window and ideally give them some time in next
> > before.
> 
> Thanks, these seem fine to me, and Lorenzo normally takes care of
> drivers/pci/controller/.  Lorenzo, if it's easier to have me apply
> them, that's fine, too, just let me know.
> 
> The only tweaks I would make would be:
> 
>   PCI: j721e: Convert to platform remove callback returning void
>   PCI: dwc: Convert to platform remove callback returning void
> 
> to match the git history.

Hello Bjorn, it should be expected that other changes for PCIe drivers
sent by other people which were sent to the list before this patch
series and are still waiting for the review (because are without
comments), would be processed before and patches sent later.

Also I would like to point out that in past I have sent fixes for PCIe
mvebu driver, which is currently in the broken state. And this is also
on waiting on the list.

Re: [PATCH 00/15] PCI: Convert to platform remove callback returning void

[Bjorn Helgaas]

On Sun, Jun 11, 2023 at 03:24:23PM +0200, Pali Rohár wrote:
> On Friday 02 June 2023 16:37:34 Bjorn Helgaas wrote:
> > On Tue, May 30, 2023 at 04:07:42PM +0200, Uwe Kleine-König wrote:
> > > On Tue, Mar 21, 2023 at 08:31:53PM +0100, Uwe Kleine-König wrote:
> > > > this series adapts the platform drivers below drivers/pci to
> > > > use the .remove_new() callback. Compared to the traditional
> > > > .remove() callback .remove_new() returns no value. This is a
> > > > good thing because the driver core doesn't (and cannot) cope
> > > > for errors during remove. The only effect of a non-zero return
> > > > value in .remove() is that the driver core emits a warning.
> > > > The device is removed anyhow and an early return from
> > > > .remove() usually yields a resource leak.
> ...

> Hello Bjorn, it should be expected that other changes for PCIe drivers
> sent by other people which were sent to the list before this patch
> series and are still waiting for the review (because are without
> comments), would be processed before and patches sent later.

I don't think it's necessary to delay simple, easily-reviewed changes
behind more complicated ones.

> Also I would like to point out that in past I have sent fixes for PCIe
> mvebu driver, which is currently in the broken state. And this is also
> on waiting on the list.

Thanks for this reminder.  Would you mind reposting them?  I poked
around in patchwork and I must be doing something wrong because I
can't find *any* patches from you, though obviously there are many.

If you repost them at least we'll know unambiguously what is on the
table.

Bjorn

Re: [PATCH 00/15] PCI: Convert to platform remove callback returning void

[Pali Rohár]

On Monday 12 June 2023 11:19:27 Bjorn Helgaas wrote:
> On Sun, Jun 11, 2023 at 03:24:23PM +0200, Pali Rohár wrote:
> > On Friday 02 June 2023 16:37:34 Bjorn Helgaas wrote:
> > > On Tue, May 30, 2023 at 04:07:42PM +0200, Uwe Kleine-König wrote:
> > > > On Tue, Mar 21, 2023 at 08:31:53PM +0100, Uwe Kleine-König wrote:
> > > > > this series adapts the platform drivers below drivers/pci to
> > > > > use the .remove_new() callback. Compared to the traditional
> > > > > .remove() callback .remove_new() returns no value. This is a
> > > > > good thing because the driver core doesn't (and cannot) cope
> > > > > for errors during remove. The only effect of a non-zero return
> > > > > value in .remove() is that the driver core emits a warning.
> > > > > The device is removed anyhow and an early return from
> > > > > .remove() usually yields a resource leak.
> > ...
> 
> > Hello Bjorn, it should be expected that other changes for PCIe drivers
> > sent by other people which were sent to the list before this patch
> > series and are still waiting for the review (because are without
> > comments), would be processed before and patches sent later.
> 
> I don't think it's necessary to delay simple, easily-reviewed changes
> behind more complicated ones.
> 
> > Also I would like to point out that in past I have sent fixes for PCIe
> > mvebu driver, which is currently in the broken state. And this is also
> > on waiting on the list.
> 
> Thanks for this reminder.  Would you mind reposting them?  I poked
> around in patchwork and I must be doing something wrong because I
> can't find *any* patches from you, though obviously there are many.
> 
> If you repost them at least we'll know unambiguously what is on the
> table.
> 
> Bjorn

Well, my patches I reposted more times. And some were also reposted by
other people. I do not know if they are in patchwork, but they are in
email archive. For example last repost of aardvark patches are here:
https://lore.kernel.org/linux-pci/20220927141926.8895-1-kabel@kernel.org/
And some other aardvark are also here:
https://lore.kernel.org/linux-pci/20220711120626.11492-1-pali@kernel.org/

Re: [PATCH 00/15] PCI: Convert to platform remove callback returning void

[Krzysztof Wilczyński]

Hello,

> this series adapts the platform drivers below drivers/pci to use the
> .remove_new() callback. Compared to the traditional .remove() callback
> .remove_new() returns no value. This is a good thing because the driver core
> doesn't (and cannot) cope for errors during remove. The only effect of a
> non-zero return value in .remove() is that the driver core emits a warning. The
> device is removed anyhow and an early return from .remove() usually yields a
> resource leak.
> 
> By changing the remove callback to return void driver authors cannot
> reasonably assume any more that there is some kind of cleanup later.
> 
> All drivers were easy to convert as they all returned zero in their
> remove callback. Only for iproc the conversion wasn't trivial, the other
> were converted using coccinelle.
> 
> There are no interdependencies between these patches. So even if there
> are some concerns for individual patches, I ask you to apply the
> remaining set. Then I only have to care for the review feedback of the
> refused patches. (Having said that I don't expect any serious objection,
> just things like squashing or separating patches, or maybe I picked a
> wrong subject prefix.)

Applied to controller/remove-void-callbacks, thank you!

[01/15] PCI: aardvark: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/b11c76db97e7
[02/15] PCI: altera: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/3a610560aa4f
[03/15] PCI: altera-msi: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/60d03f70455c
[04/15] PCI: brcmstb: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/b169c576ad0c
[05/15] PCI: j721e: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/c86f4bd6008e
[06/15] PCI: dwc: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/221879c98698
[07/15] PCI: hisi-error: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/9a285fbbb591
[08/15] PCI: iproc: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/6f1c0a046048
[09/15] PCI: mediatek: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/5e0005409427
[10/15] PCI: mediatek-gen3: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/22626c46bb32
[11/15] PCI: mt7621: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/8c47ac2a66c4
[12/15] PCI: mvebu: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/4c3bc1b41b8f
[13/15] PCI: rockchip-host: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/2998efcd8e73
[14/15] PCI: tegra: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/c7fd95cda648
[15/15] PCI: xgene-msi: Convert to platform remove callback returning void
        https://git.kernel.org/pci/pci/c/afbb9130d2bf

	Krzysztof

Re: [PATCH 00/15] PCI: Convert to platform remove callback returning void

[Krzysztof Wilczyński]

Hello,

[...]
> > > These patches wait for application for quite some time now. They apply
> > > just fine to v6.4-rc1 and next/master. Would be great to get them in
> > > during the next merge window and ideally give them some time in next
> > > before.
> > 
> > Thanks, these seem fine to me, and Lorenzo normally takes care of
> > drivers/pci/controller/.  Lorenzo, if it's easier to have me apply
> > them, that's fine, too, just let me know.
> > 
> > The only tweaks I would make would be:
> > 
> >   PCI: j721e: Convert to platform remove callback returning void
> >   PCI: dwc: Convert to platform remove callback returning void
> 
> If it's easier for you (or Lorenzo) I can resend with these tweaks.
> Otherwise if these are adapted when applying them, that's fine for me,
> too. Just tell me if I should do anything here.

I took the entire series and tweaked the subject lines as Bjorn requested.

Thank you!

	Krzysztof