Re: [PATCH][RESEND] wistron_btns needs executable BIOS image

Re: [PATCH][RESEND] wistron_btns needs executable BIOS image

[Eric Wong]

Jakub Bogusz <qboosh@pld-linux.org> wrote:
> Let's try once again... (take 3)
> 
> First time I sent this patch when I prepared it in 2013 for Linux 3.12,
> the second time after update for Linux 4.12 in 2017...
> And it still applies to Linux 4.20.

Adding Dmitry Torokhov and linux-input to Cc:
(that's what ./scripts/get_maintainer.pl says)

> Actual description in forwarded message.
> 
> 
> -- 
> Jakub Bogusz    http://qboosh.pl/

> From: Jakub Bogusz <qboosh@pld-linux.org>
> To: linux-kernel@vger.kernel.org, mitr@volny.cz
> Subject: [PATCH] wistron_btns needs executable BIOS image
> 
> Hello,
> 
> This patch (originally agains 3.1x, now I updated include to build
> against 4.12.x) fixes winstron_btns module issue with calling BIOS
> functions in non-executable memory.
> 
> Tested (on Linux 3.10.x and few later versions) on F-S Amilo 8210 laptop.
> 
> 
> -- 
> Jakub Bogusz    http://qboosh.pl/

> wistron_btns needs executable BIOS image.
> 
> Signed-off-by: Jakub Bogusz <qboosh@pld-linux.org>
> 
> --- linux-4.12/drivers/input/misc/wistron_btns.c.orig	2013-11-16 09:05:55.612742472 +0100
> +++ linux-4.12/drivers/input/misc/wistron_btns.c	2013-11-16 09:24:37.356028732 +0100
> @@ -33,6 +33,7 @@
>  #include <linux/types.h>
>  #include <linux/platform_device.h>
>  #include <linux/leds.h>
> +#include <asm/set_memory.h>
>  
>  /* How often we poll keys - msecs */
>  #define POLL_INTERVAL_DEFAULT	500 /* when idle */
> @@ -124,6 +125,7 @@
>  	if (entry_point >= 0xF0000) {
>  		bios_code_map_base = base;
>  		bios_entry_point = bios_code_map_base + (entry_point & 0xFFFF);
> +		set_memory_x((unsigned long)bios_code_map_base, 0x10000 >> PAGE_SHIFT);
>  	} else {
>  		iounmap(base);
>  		bios_code_map_base = ioremap(entry_point & ~0x3FFF, 0x4000);
> @@ -134,6 +136,7 @@
>  			goto err;
>  		}
>  		bios_entry_point = bios_code_map_base + (entry_point & 0x3FFF);
> +		set_memory_x((unsigned long)bios_code_map_base, 0x4000 >> PAGE_SHIFT);
>  	}
>  	/* The Windows driver maps 0x10000 bytes, we keep only one page... */
>  	bios_data_map_base = ioremap(0x400, 0xc00);

Re: [PATCH][RESEND] wistron_btns needs executable BIOS image

[Dmitry Torokhov]

On Fri, Jan 25, 2019 at 1:13 PM Eric Wong <e@80x24.org> wrote:
>
> Jakub Bogusz <qboosh@pld-linux.org> wrote:
> > Let's try once again... (take 3)
> >
> > First time I sent this patch when I prepared it in 2013 for Linux 3.12,
> > the second time after update for Linux 4.12 in 2017...
> > And it still applies to Linux 4.20.
>
> Adding Dmitry Torokhov and linux-input to Cc:
> (that's what ./scripts/get_maintainer.pl says)
>
> > Actual description in forwarded message.
> >
> >
> > --
> > Jakub Bogusz    http://qboosh.pl/
>
> > From: Jakub Bogusz <qboosh@pld-linux.org>
> > To: linux-kernel@vger.kernel.org, mitr@volny.cz
> > Subject: [PATCH] wistron_btns needs executable BIOS image
> >
> > Hello,
> >
> > This patch (originally agains 3.1x, now I updated include to build
> > against 4.12.x) fixes winstron_btns module issue with calling BIOS
> > functions in non-executable memory.
> >
> > Tested (on Linux 3.10.x and few later versions) on F-S Amilo 8210 laptop.
> >
> >
> > --
> > Jakub Bogusz    http://qboosh.pl/
>
> > wistron_btns needs executable BIOS image.
> >
> > Signed-off-by: Jakub Bogusz <qboosh@pld-linux.org>
> >
> > --- linux-4.12/drivers/input/misc/wistron_btns.c.orig 2013-11-16 09:05:55.612742472 +0100
> > +++ linux-4.12/drivers/input/misc/wistron_btns.c      2013-11-16 09:24:37.356028732 +0100
> > @@ -33,6 +33,7 @@
> >  #include <linux/types.h>
> >  #include <linux/platform_device.h>
> >  #include <linux/leds.h>
> > +#include <asm/set_memory.h>
> >
> >  /* How often we poll keys - msecs */
> >  #define POLL_INTERVAL_DEFAULT        500 /* when idle */
> > @@ -124,6 +125,7 @@
> >       if (entry_point >= 0xF0000) {
> >               bios_code_map_base = base;
> >               bios_entry_point = bios_code_map_base + (entry_point & 0xFFFF);
> > +             set_memory_x((unsigned long)bios_code_map_base, 0x10000 >> PAGE_SHIFT);
> >       } else {
> >               iounmap(base);
> >               bios_code_map_base = ioremap(entry_point & ~0x3FFF, 0x4000);
> > @@ -134,6 +136,7 @@
> >                       goto err;
> >               }
> >               bios_entry_point = bios_code_map_base + (entry_point & 0x3FFF);
> > +             set_memory_x((unsigned long)bios_code_map_base, 0x4000 >> PAGE_SHIFT);
> >       }
> >       /* The Windows driver maps 0x10000 bytes, we keep only one page... */
> >       bios_data_map_base = ioremap(0x400, 0xc00);

It is really weird have __iomem region to have executable bit set. I
wonder if, along with the setting memory to X, we should not change
that ioremap() to memremap()?

Thanks.

-- 
Dmitry

Re: [PATCH][RESEND] wistron_btns needs executable BIOS image

[Jakub Bogusz]

On Fri, Jan 25, 2019 at 01:41:47PM -0800, Dmitry Torokhov wrote:
> On Fri, Jan 25, 2019 at 1:13 PM Eric Wong <e@80x24.org> wrote:
> >
> > Jakub Bogusz <qboosh@pld-linux.org> wrote:
> > > Let's try once again... (take 3)
> > >
> > > First time I sent this patch when I prepared it in 2013 for Linux 3.12,
> > > the second time after update for Linux 4.12 in 2017...
> > > And it still applies to Linux 4.20.
> >
> > Adding Dmitry Torokhov and linux-input to Cc:
> > (that's what ./scripts/get_maintainer.pl says)

Thanks. I was aware only of MAINTAINERS file.

[...]
> > > wistron_btns needs executable BIOS image.
> > >
> > > Signed-off-by: Jakub Bogusz <qboosh@pld-linux.org>
> > >
> > > --- linux-4.12/drivers/input/misc/wistron_btns.c.orig 2013-11-16 09:05:55.612742472 +0100
> > > +++ linux-4.12/drivers/input/misc/wistron_btns.c      2013-11-16 09:24:37.356028732 +0100
> > > @@ -33,6 +33,7 @@
> > >  #include <linux/types.h>
> > >  #include <linux/platform_device.h>
> > >  #include <linux/leds.h>
> > > +#include <asm/set_memory.h>
> > >
> > >  /* How often we poll keys - msecs */
> > >  #define POLL_INTERVAL_DEFAULT        500 /* when idle */
> > > @@ -124,6 +125,7 @@
> > >       if (entry_point >= 0xF0000) {
> > >               bios_code_map_base = base;
> > >               bios_entry_point = bios_code_map_base + (entry_point & 0xFFFF);
> > > +             set_memory_x((unsigned long)bios_code_map_base, 0x10000 >> PAGE_SHIFT);
> > >       } else {
> > >               iounmap(base);
> > >               bios_code_map_base = ioremap(entry_point & ~0x3FFF, 0x4000);
> > > @@ -134,6 +136,7 @@
> > >                       goto err;
> > >               }
> > >               bios_entry_point = bios_code_map_base + (entry_point & 0x3FFF);
> > > +             set_memory_x((unsigned long)bios_code_map_base, 0x4000 >> PAGE_SHIFT);
> > >       }
> > >       /* The Windows driver maps 0x10000 bytes, we keep only one page... */
> > >       bios_data_map_base = ioremap(0x400, 0xc00);
> 
> It is really weird have __iomem region to have executable bit set. I
> wonder if, along with the setting memory to X, we should not change
> that ioremap() to memremap()?

Maybe... this is area of BIOS ROM, not I/O. But I'm not sure about
cacheability.

Looking at recent kernel code (4.19 to be exact) ioremap ends in
__ioremap_caller with _PAGE_CACHE_MODE_UC_MINUS, while memremap would end
in __ioremap_caller with e.g. (depending on flags) _PAGE_CACHE_MODE_WT.

Unfortunately now I have rare access to this hardware (FS Amilo) to
test.


-- 
Jakub Bogusz    http://qboosh.pl/