# \[PATCH\] video: fbdev: uvesafb: use CONFIG\_X86\_PAE surround \_PAGE\_NX check code - Subsystems: FRAMEBUFFER LAYER, UVESAFB DRIVER, THE REST - Maintainers: Helge Deller, Michal Januszewski, Linus Torvalds - List: linux-fbdev - 3 messages, 2 authors - Page: https://ratatoskr.run/linux-fbdev/2014/03/4211988/t ## \[PATCH\] video: fbdev: uvesafb: use CONFIG\_X86\_PAE surround \_PAGE\_NX check code - From: Wang YanQing \ - Date: 2014-03-05 16:17:20 - Message: https://ratatoskr.run/linux-fbdev/2014/03/4211988.md ```text Because _PAGE_NX check will always false when we don't define CONFIG_X86_PAE for CONFIG_X86_32, so use CONFIG_X86_PAE surround the check code. Although I believe "smart" compile will optimize out and generate the same code, but use CONFIG_X86_PAE surround check code will clear it and prohibit warning from static source code analyze tool. [ This patch fix warning report by fengguang.wu@intel.com "drivers/video/fbdev/uvesafb.c:816 uvesafb_vbe_init() warn: bitwise AND condition is false here" ] Signed-off-by: Wang YanQing --- drivers/video/fbdev/uvesafb.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/video/fbdev/uvesafb.c b/drivers/video/fbdev/uvesafb.c index 509d452..102858c 100644 --- a/drivers/video/fbdev/uvesafb.c +++ b/drivers/video/fbdev/uvesafb.c @@ -813,6 +813,7 @@ static int uvesafb_vbe_init(struct fb_info *info) par->ypan = ypan; if (par->pmi_setpal || par->ypan) { +#ifdef CONFIG_X86_PAE if (__supported_pte_mask & _PAGE_NX) { par->pmi_setpal = par->ypan = 0; printk(KERN_WARNING "uvesafb: NX protection is active, " @@ -820,6 +821,9 @@ static int uvesafb_vbe_init(struct fb_info *info) } else { uvesafb_vbe_getpmi(task, par); } +#else + uvesafb_vbe_getpmi(task, par); +#endif } #else /* The protected mode interface is not available on non-x86. */ -- 1.8.3.4.8.g69490f3.dirty ``` ## Re: \[PATCH\] video: fbdev: uvesafb: use CONFIG\_X86\_PAE surround \_PAGE\_NX check code - From: Tomi Valkeinen \ - Date: 2014-03-06 09:56:48 - Message: https://ratatoskr.run/linux-fbdev/2014/03/4212535.md ```text On 05/03/14 18:16, Wang YanQing wrote: > Because _PAGE_NX check will always false when we don't define > CONFIG_X86_PAE for CONFIG_X86_32, so use CONFIG_X86_PAE surround > the check code. > > Although I believe "smart" compile will optimize out and generate > the same code, but use CONFIG_X86_PAE surround check code will > clear it and prohibit warning from static source code analyze tool. > > [ This patch fix warning report by fengguang.wu@intel.com > "drivers/video/fbdev/uvesafb.c:816 uvesafb_vbe_init() > warn: bitwise AND condition is false here" ] > > Signed-off-by: Wang YanQing > --- > drivers/video/fbdev/uvesafb.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/drivers/video/fbdev/uvesafb.c b/drivers/video/fbdev/uvesafb.c > index 509d452..102858c 100644 > --- a/drivers/video/fbdev/uvesafb.c > +++ b/drivers/video/fbdev/uvesafb.c > @@ -813,6 +813,7 @@ static int uvesafb_vbe_init(struct fb_info *info) > par->ypan = ypan; > > if (par->pmi_setpal || par->ypan) { > +#ifdef CONFIG_X86_PAE > if (__supported_pte_mask & _PAGE_NX) { > par->pmi_setpal = par->ypan = 0; > printk(KERN_WARNING "uvesafb: NX protection is active, " > @@ -820,6 +821,9 @@ static int uvesafb_vbe_init(struct fb_info *info) > } else { > uvesafb_vbe_getpmi(task, par); > } > +#else > + uvesafb_vbe_getpmi(task, par); > +#endif > } I don't like this, I think this makes the code more messy, just to avoid that warning. And it might even be buggy. For your patch to work correctly, you need to know the internals of _PAGE_NX, i.e. that when CONFIG_X86_PAE is not defined, _PAGE_NX is 0. But the driver should not depend on things like that. If _PAGE_NX is changed later, the driver will not work correctly. Tomi ``` ## Re: \[PATCH\] video: fbdev: uvesafb: use CONFIG\_X86\_PAE surround \_PAGE\_NX check code - From: Wang YanQing \ - Date: 2014-03-06 10:34:48 - Message: https://ratatoskr.run/linux-fbdev/2014/03/4212569.md ```text On Thu, Mar 06, 2014 at 11:56:37AM +0200, Tomi Valkeinen wrote: > I don't like this, I think this makes the code more messy, just to avoid > that warning. > > And it might even be buggy. For your patch to work correctly, you need > to know the internals of _PAGE_NX, i.e. that when CONFIG_X86_PAE is not > defined, _PAGE_NX is 0. But the driver should not depend on things like > that. If _PAGE_NX is changed later, the driver will not work correctly. > Very right! Thanks. ```