what is linux-gate.so

[Gavin Li]

Hi,

 

When I do ldd a.out, it depends linux-gate.so.1, what is it? I can't find it
in my whole system.

[Dave Jones]

On Wed, Dec 14, 2005 at 12:35:01PM -0500, Gavin Li wrote:

>  When I do ldd a.out, it depends linux-gate.so.1, what is it? I
> can't find it
>  in my whole system. 
It's the vdso where vsyscalls live.  The kernel provides it.

		Dave

[Chitlesh GOORAH]

On 12/14/05, Dave Jones <davej(a)redhat.com&gt; wrote:
>  On Wed, Dec 14, 2005 at 12:35:01PM -0500, Gavin Li wrote:
>
>   > When I do ldd a.out, it depends linux-gate.so.1, what is it? I can't find it
>   > in my whole system.
>
>  It's the vdso where vsyscalls live.  The kernel provides it.
>
>                  Dave
>  
I have seen everywhere on internet  linux-gate.so.1 =>  (0xffffe000)

but why is my    linux-gate.so.1 =>  (0x00732000)

?
>  -- 


-- 
http://clunixchit.blogspot.com

[Dave Jones]

On Thu, Dec 28, 2006 at 07:14:39PM +0100, Chitlesh GOORAH wrote:
>  On 12/14/05, Dave Jones <davej(a)redhat.com&gt; wrote:
>  > On Wed, Dec 14, 2005 at 12:35:01PM -0500, Gavin Li wrote:
>  >
>  >  > When I do ldd a.out, it depends linux-gate.so.1, what is it? I can't find
> it
>  >  > in my whole system.
>  >
>  > It's the vdso where vsyscalls live.  The kernel provides it.
>  >
>  >                 Dave
>  >
>  
>  I have seen everywhere on internet  linux-gate.so.1 =>  (0xffffe000)
>  
>  but why is my    linux-gate.so.1 =>  (0x00732000) 
Address space randomization.  Exec-shield does some things that
the upstream kernel doesn't, which could explain the disparity.
(or it could be that the examples you saw are from older kernels
 which didn't have randomization).

		Dave

-- 
http://www.codemonkey.org.uk

[Tom Horsley]

On Thu, 28 Dec 2006 19:14:39 +0100
"Chitlesh GOORAH" <chitlesh(a)fedoraproject.org&gt; wrote:

>  I have seen everywhere on internet  linux-gate.so.1 => 
> (0xffffe000)
>  
>  but why is my    linux-gate.so.1 =>  (0x00732000) 
Beacuse you have a new and improved kernel that randomizes
the address space to give the folks who implemented the address
space randomizer a false sense of security about preventing
buffer overrun exploits while making it impossible to actually
debug any real buffer overrun problems by making it impossible
to ever reproduce the same program behavior :-).

(Meanwhile a completely different group of fanatics are spending
all their time trying to make programs load 1 nanosecond faster
by having the system spend hours pre-relocating libraries so they
will load faster at a particular address they will never actually
load at because the kernel randomizes the load addresses of everything.
Too bad these folks never talk to each other :-).

If you run some more program and check their /proc/pid/maps
file, you'll find linux-gate.so.1 in lots of different places.