Because memory ranges in mem->ranges are stored in ascending order, when we
detect `p_end < start`, we can break the for loop early, as the subsequent
memory ranges must also be outside the range we are looking for.

Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
---
Hi Andrew,

Patch "[PATCH 2/2] crash_core: fix out-of-bounds access check in
crash_exclude_mem_range()" can be ignored, use this patch instead.

 kernel/crash_core.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/kernel/crash_core.c b/kernel/crash_core.c
index 9a219a918638..d425c4a106cd 100644
--- a/kernel/crash_core.c
+++ b/kernel/crash_core.c
@@ -575,9 +575,12 @@ int crash_exclude_mem_range(struct crash_mem *mem,
 		p_start = mstart;
 		p_end = mend;
 
-		if (p_start > end || p_end < start)
+		if (p_start > end)
 			continue;
 
+		if (p_end < start)
+			break;
+
 		/* Truncate any area outside of range */
 		if (p_start < start)
 			p_start = start;
-- 
2.43.0

On Wed, 20 Dec 2023 00:34:18 +0800 Yuntao Wang <ytcoode@gmail.com> wrote:

> Because memory ranges in mem->ranges are stored in ascending order, when we
> detect `p_end < start`, we can break the for loop early, as the subsequent
> memory ranges must also be outside the range we are looking for.
> 
> Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
> ---
> Hi Andrew,
> 
> Patch "[PATCH 2/2] crash_core: fix out-of-bounds access check in
> crash_exclude_mem_range()" can be ignored, use this patch instead.
> 

Some reviewer input on this would be helpful please?

> --- a/kernel/crash_core.c
> +++ b/kernel/crash_core.c
> @@ -575,9 +575,12 @@ int crash_exclude_mem_range(struct crash_mem *mem,
>  		p_start = mstart;
>  		p_end = mend;
>  
> -		if (p_start > end || p_end < start)
> +		if (p_start > end)
>  			continue;
>  
> +		if (p_end < start)
> +			break;
> +
>  		/* Truncate any area outside of range */
>  		if (p_start < start)
>  			p_start = start;
> -- 
> 2.43.0

On 12/29/23 at 12:10pm, Andrew Morton wrote:
> On Wed, 20 Dec 2023 00:34:18 +0800 Yuntao Wang <ytcoode@gmail.com> wrote:
> 
> > Because memory ranges in mem->ranges are stored in ascending order, when we
> > detect `p_end < start`, we can break the for loop early, as the subsequent
> > memory ranges must also be outside the range we are looking for.
> > 
> > Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
> > ---
> > Hi Andrew,
> > 
> > Patch "[PATCH 2/2] crash_core: fix out-of-bounds access check in
> > crash_exclude_mem_range()" can be ignored, use this patch instead.
> > 
> 
> Some reviewer input on this would be helpful please?


I suggested this in below discussion thread:
https://lore.kernel.org/all/ZYEOshALGbDKwSdc@MiWiFi-R3L-srv/T/#u

So it would be good if squashing this into patch 3 of another patch
thread you are asking:
[PATCH 3/3] crash_core: fix and simplify the logic of crash_exclude_mem_range()

And I would suggest withdrawing Yuntao's below patch on your
mm-nonmm-unstable branch.

961c69e9f1bf x86/crash: fix potential cmem->ranges array overflow

Becase there's better one to fix the potential oob from fuqiang,
although fuqiang need improve his patch log.

[PATCH v3] x86/kexec: fix potential cmem->ranges out of bounds
https://lore.kernel.org/all/20231222121855.148215-1-fuqiang.wang@easystack.cn/T/#u

> 
> > --- a/kernel/crash_core.c
> > +++ b/kernel/crash_core.c
> > @@ -575,9 +575,12 @@ int crash_exclude_mem_range(struct crash_mem *mem,
> >  		p_start = mstart;
> >  		p_end = mend;
> >  
> > -		if (p_start > end || p_end < start)
> > +		if (p_start > end)
> >  			continue;
> >  
> > +		if (p_end < start)
> > +			break;
> > +
> >  		/* Truncate any area outside of range */
> >  		if (p_start < start)
> >  			p_start = start;
> > -- 
> > 2.43.0
>

On Sat, 30 Dec 2023 18:28:06 +0800, Baoquan He <bhe@redhat.com> wrote:

> On 12/29/23 at 12:10pm, Andrew Morton wrote:
> > On Wed, 20 Dec 2023 00:34:18 +0800 Yuntao Wang <ytcoode@gmail.com> wrote:
> > 
> > > Because memory ranges in mem->ranges are stored in ascending order, when we
> > > detect `p_end < start`, we can break the for loop early, as the subsequent
> > > memory ranges must also be outside the range we are looking for.
> > > 
> > > Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
> > > ---
> > > Hi Andrew,
> > > 
> > > Patch "[PATCH 2/2] crash_core: fix out-of-bounds access check in
> > > crash_exclude_mem_range()" can be ignored, use this patch instead.
> > > 
> > 
> > Some reviewer input on this would be helpful please?
> 
> 
> I suggested this in below discussion thread:
> https://lore.kernel.org/all/ZYEOshALGbDKwSdc@MiWiFi-R3L-srv/T/#u
> 
> So it would be good if squashing this into patch 3 of another patch
> thread you are asking:
> [PATCH 3/3] crash_core: fix and simplify the logic of crash_exclude_mem_range()
>

Hi all,

I've squashed this patch into the patch:

[PATCH 3/3] crash_core: fix and simplify the logic of crash_exclude_mem_range()

The link to the new patch is:

https://lore.kernel.org/lkml/20240102144905.110047-1-ytcoode@gmail.com/t/#m255d0d26148f2b384f6b7ab77eb38edf3f1bc0df

> And I would suggest withdrawing Yuntao's below patch on your
> mm-nonmm-unstable branch.
> 
> 961c69e9f1bf x86/crash: fix potential cmem->ranges array overflow
> 
> Becase there's better one to fix the potential oob from fuqiang,
> although fuqiang need improve his patch log.
> 
> [PATCH v3] x86/kexec: fix potential cmem->ranges out of bounds
> https://lore.kernel.org/all/20231222121855.148215-1-fuqiang.wang@easystack.cn/T/#u
>

I'm okay with that.

> > 
> > > --- a/kernel/crash_core.c
> > > +++ b/kernel/crash_core.c
> > > @@ -575,9 +575,12 @@ int crash_exclude_mem_range(struct crash_mem *mem,
> > >  		p_start = mstart;
> > >  		p_end = mend;
> > >  
> > > -		if (p_start > end || p_end < start)
> > > +		if (p_start > end)
> > >  			continue;
> > >  
> > > +		if (p_end < start)
> > > +			break;
> > > +
> > >  		/* Truncate any area outside of range */
> > >  		if (p_start < start)
> > >  			p_start = start;
> > > -- 
> > > 2.43.0
> >