linux-kernel - Re: coccinelle: How to remove a return at the end of a void function?

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: <be39f7d-31ab-24c-61c5-6daabdef9267@inria.fr>
Date:   Mon, 26 Dec 2022 17:02:35 +0100 (CET)
From:   Julia Lawall <julia.lawall@...ia.fr>
To:     Uwe Kleine-König 
        <u.kleine-koenig@...gutronix.de>
cc:     kernel@...gutronix.de, Nicolas Palix <nicolas.palix@...g.fr>,
        linux-kernel@...r.kernel.org, cocci@...ia.fr
Subject: Re: coccinelle: How to remove a return at the end of a void
 function?



On Sun, 25 Dec 2022, Uwe Kleine-König wrote:

> Hello Julia,
>
> first of all thanks for your quick answer.
>
> On Sat, Dec 24, 2022 at 01:28:04PM +0100, Julia Lawall wrote:
> > On Sat, 24 Dec 2022, Uwe Kleine-König wrote:
> > > A simplified spatch looks as follows:
> > >
> > > -------->8--------
> > > virtual patch
> > >
> > > @p1@
> > > identifier pdev;
> > > @@
> > > -int
> > > +void
> > >  rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev) {
> > >  <...
> > > -return 0;
> > > +return;
> > >  ...>
> > >  }
> > > -------->8--------
> > >
> > > This results in:
> > >
> > > -------->8--------
> > > diff -u -p a/drivers/mmc/host/rtsx_usb_sdmmc.c b/drivers/mmc/host/rtsx_usb_sdmmc.c
> > > --- a/drivers/mmc/host/rtsx_usb_sdmmc.c
> > > +++ b/drivers/mmc/host/rtsx_usb_sdmmc.c
> > > @@ -1379,13 +1379,13 @@ static int rtsx_usb_sdmmc_drv_probe(stru
> > >         return 0;
> > >  }
> > >
> > > -static int rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev)
> > > +static void rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev)
> > >  {
> > >         struct rtsx_usb_sdmmc *host = platform_get_drvdata(pdev);
> > >         struct mmc_host *mmc;
> > >
> > >         if (!host)
> > > -               return 0;
> > > +               return;
> > >
> > >         mmc = host->mmc;
> > >         host->host_removal = true;
> > > @@ -1416,7 +1416,7 @@ static int rtsx_usb_sdmmc_drv_remove(str
> > >         dev_dbg(&(pdev->dev),
> > >                 ": Realtek USB SD/MMC module has been removed\n");
> > >
> > > -       return 0;
> > > +       return;
> > >  }
> > >
> > >  #ifdef CONFIG_PM
> > > -------->8--------
> > >
> > > which is as intended. Now I want to remove the useless "return;" at the
> > > end of the function, however adding
> > >
> > > -------->8--------
> > > @p2 depends on p1@
> > > identifier pdev;
> > > @@
> > >  void rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev) {
> > >  ...
> > > -return;
> > >  }
> > > -------->8--------
> > >
> > > to the spatch doesn't (only) do the intended:
> >
> > The problem is that Coccinelle is following the control-flow through the
> > function, and all of the returns are at the end of a control.flow path.
> > The simple, hacky solution is to change the return;s into some function
> > call Return();, then do like the above for Return(); and then change the
> > Return();s back to return;s
>
> OK, I tried, but somehow coccinelle refuse to work after I introduced
> Return(), even replacing them by return; doesn't work:
>
> -------->8--------
> virtual patch
>
> @p1@
> identifier pdev;
> @@
> -int
> +void
>  rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev) {
>  ...
> -return 0;
> +Return();
>  ...
>  }
>
> @p2@
> identifier pdev;
> @@
>  void rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev) {
>  ...
> -Return();
> +return;
>  ...
>  }

The problem is that a control-flow path at this point can have multiple
calls to Return();  You pattern only matches when every control-flow path
through the code has exactly one Return().

You should have one rule that removes the final Return();

 @p2@
 identifier pdev;
 @@
  void rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev) {
  ...
 -Return();
  }

Then another rule to remove the others:

 @p2@
 identifier pdev;
 @@
  void rtsx_usb_sdmmc_drv_remove(struct platform_device *pdev) {
  <...
 -Return();
 +return;
  ...>
  }

julia

> -------->8--------
>
> results in
>
> -------->8--------
> $ /usr/bin/spatch --debug -D patch --very-quiet --cocci-file scripts/coccinelle/api/test.cocci --patch . --dir drivers/mmc/host/rtsx_usb_sdmmc.c -I ./arch/x86/include -I ./arch/x86/include/generated -I ./include -I ./arch/x86/include/uapi -I ./arch/x86/include/generated/uapi -I ./include/uapi -I ./include/generated/uapi --include ./include/linux/compiler-version.h --include ./include/linux/kconfig.h --jobs 4 --chunksize 1
> diff -u -p a/drivers/mmc/host/rtsx_usb_sdmmc.c b/drivers/mmc/host/rtsx_usb_sdmmc.c
> --- a/drivers/mmc/host/rtsx_usb_sdmmc.c
> +++ b/drivers/mmc/host/rtsx_usb_sdmmc.c
> @@ -1385,7 +1385,7 @@ static int rtsx_usb_sdmmc_drv_remove(str
>  	struct mmc_host *mmc;
>
>  	if (!host)
> -		return 0;
> +		Return();
>
>  	mmc = host->mmc;
>  	host->host_removal = true;
> @@ -1416,7 +1416,7 @@ static int rtsx_usb_sdmmc_drv_remove(str
>  	dev_dbg(&(pdev->dev),
>  		": Realtek USB SD/MMC module has been removed\n");
>
> -	return 0;
> +	Return();
>  }
>
>  #ifdef CONFIG_PM
> -------->8--------
>
> Adding --debug doesn't give any hints.
>
> (And if I add another hunk inbeetween removing Return at the end of the
> function there is no effect either.)
>
> Do I need to split that in two spatches to make coccinelle cooperate?
>
> (If it matters, this is coccinelle as shipped by Debian, Version
> 1.1.1.deb-2)
>
> Best regards
> Uwe
>
> --
> Pengutronix e.K.                           | Uwe Kleine-König            |
> Industrial Linux Solutions                 | https://www.pengutronix.de/ |
>