Re: bug in sproc_syscall.c?

To: Gopalakrishnan Suresh <gsuresh@research.bell-labs.com>
Subject: Re: bug in sproc_syscall.c?
From: k-abe@media.osaka-cu.ac.jp
Date: Mon, 01 Jul 2002 11:51:17 +0900
Cc: <oskit-users@fast.cs.utah.edu>
In-Reply-To: <Pine.SOL.4.33.0206281819530.3201-100000@aura.research.bell-labs.com>
Organization: Media Center, Osaka City University
References: <Pine.SOL.4.33.0206271617570.15952-100000@aura.research.bell-labs.com> <Pine.SOL.4.33.0206281819530.3201-100000@aura.research.bell-labs.com>
Sender: owner-oskit-users@fast.cs.utah.edu
User-Agent: Wanderlust/2.9.10 (Unchained Melody) SEMI/1.14.3 (Ushinoya) SLIM/1.14.7 (=?ISO-2022-JP?B?GyRCPHIwZjpMTD4bKEI=?=) APEL/10.3 MULE XEmacs/21.1 (patch 14) (Cuyahoga Valley) (i386-redhat-linux)

At Fri, 28 Jun 2002 18:25:33 -0400 (EDT),
Gopalakrishnan Suresh wrote:
> 
> 
> In uvm/sproc/x86/sproc_syscall.c:oskit_sproc_syscall_handler()
> 
>     /* Dispatch the system call */
>     error = (*func)(sthread, (void*)arg_kv, rval);
> 
> Shouldn't this be
> 
>     error = (*func)(sthread, (void*)args, rval);

You are right.  Someone pointed out this bug some months ago but I
forgot to make this public.  

> Or maybe it doesn't matter - because it is a single linear address space
> (without LDTs) and kernel has access to all of it?

Actually it does matter.  args_kv points within a user address space
which might be switched by thread context switches!

Kota

References:

Re: sproc sample kernel

From: Gopalakrishnan Suresh <gsuresh@research.bell-labs.com>

bug in sproc_syscall.c?

From: Gopalakrishnan Suresh <gsuresh@research.bell-labs.com>