linux-kernel - Re: [PATCH v2] ahci: st: Switch from CONFIG_PM_SLEEP guards to pm_sleep

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Z4EDKUb+hO0ovV2i@x1-carbon>
Date: Fri, 10 Jan 2025 12:23:21 +0100
From: Niklas Cassel <cassel@...nel.org>
To: Damien Le Moal <dlemoal@...nel.org>
Cc: Raphael Gallais-Pou <rgallaispou@...il.com>,
	Patrice Chotard <patrice.chotard@...s.st.com>,
	linux-arm-kernel@...ts.infradead.org, linux-ide@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] ahci: st: Switch from CONFIG_PM_SLEEP guards to
 pm_sleep_ptr()

On Fri, Jan 10, 2025 at 02:19:06PM +0900, Damien Le Moal wrote:
> On 1/10/25 02:54, Raphael Gallais-Pou wrote:
> > Letting the compiler remove these functions when the kernel is built
> > without CONFIG_PM_SLEEP support is simpler and less error prone than the
> > use of #ifdef based kernel configuration guards.
> > 
> > Signed-off-by: Raphael Gallais-Pou <rgallaispou@...il.com>
> > ---
> > Changes in v2:
> >   - Split serie in single patches
> >   - Remove irrelevant 'Link:' from commit log
> >   - Link to v1: https://lore.kernel.org/r/20241229-update_pm_macro-v1-4-c7d4c4856336@gmail.com
> > ---
> >  drivers/ata/ahci_st.c | 6 ++----
> >  1 file changed, 2 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/ata/ahci_st.c b/drivers/ata/ahci_st.c
> > index 6b9b4a1dfa15..4336c8a6e208 100644
> > --- a/drivers/ata/ahci_st.c
> > +++ b/drivers/ata/ahci_st.c
> > @@ -176,7 +176,6 @@ static int st_ahci_probe(struct platform_device *pdev)
> >  	return 0;
> >  }
> >  
> > -#ifdef CONFIG_PM_SLEEP
> >  static int st_ahci_suspend(struct device *dev)
> >  {
> >  	struct ata_host *host = dev_get_drvdata(dev);
> > @@ -221,9 +220,8 @@ static int st_ahci_resume(struct device *dev)
> >  
> >  	return ahci_platform_resume_host(dev);
> >  }
> > -#endif
> 
> I do not think you can remove the ifdef here. Otherwise, there is going to be a
> compilation warning when CONFIG_PM_SLEEP is not enabled. No ?

Look at the pm_sleep_ptr macro:
include/linux/pm.h:#define pm_sleep_ptr(_ptr) PTR_IF(IS_ENABLED(CONFIG_PM_SLEEP), (_ptr))

I would expect the function should be optimized out by the compiler
using dead code elimination.

Raphael, perhaps you could show the before and after output
using ./scripts/bloat-o-meter ?
(When the config is not enabled: before and after your patch.)


> 
> >  
> > -static SIMPLE_DEV_PM_OPS(st_ahci_pm_ops, st_ahci_suspend, st_ahci_resume);
> > +static DEFINE_SIMPLE_DEV_PM_OPS(st_ahci_pm_ops, st_ahci_suspend, st_ahci_resume);
> >  
> >  static const struct of_device_id st_ahci_match[] = {
> >  	{ .compatible = "st,ahci", },
> > @@ -234,7 +232,7 @@ MODULE_DEVICE_TABLE(of, st_ahci_match);
> >  static struct platform_driver st_ahci_driver = {
> >  	.driver = {
> >  		.name = DRV_NAME,
> > -		.pm = &st_ahci_pm_ops,
> > +		.pm = pm_sleep_ptr(&st_ahci_pm_ops),
> >  		.of_match_table = st_ahci_match,
> >  	},
> >  	.probe = st_ahci_probe,
> 
> 
> -- 
> Damien Le Moal
> Western Digital Research