linux emulation COMPAT_43 removal part 1

Matthew Dillon dillon at apollo.backplane.com
Sat Sep 6 13:38:40 PDT 2003


    oops, resending. I manged David's email address.

:>     base, but I could go either way.  e.g. kern_bind() instead of bind1().
:>     Definite not _syscall_bind() (ick).  But before we make that sort of
:>     change I will commit the linux patch using the current naming 
:>     convention.
:
:	Deal!  If I get some time in the next two weeks, I will convert
:	as many syscalls as I can into the split system, and then
:	collectively change their name from blah1() to kern_blah().
:
:	Regards,
:
:		- Hiten (hmp at xxxxxxxxxxxxx)

    Woa, timeout!  Coordinate with David Reese, he's been working on
    splitting the system calls as well.  There are a lot of syscalls so
    the help is appreciated, I suggest that you guys divvy up the files
    as you go.

    -

    David, I reviewed your patch and I thought I found a bug, but it turns
    out the bug is in the original API that you duplicated exactly!  
    getsockaddr() (and consequentially linux_getsockaddr()) was not checking
    for a passed length parameter that was too small.  This can result in 
    a 0-length malloc and then an attempt to assign sa->sa_len.

    Also, I looked at the FreeBSD-current code and they appear to be using
    kern_*(), so if we adopt kern_*() we will be using a comparable API.

    In anycase, don't worry about the malloc problem, I will commit a check
    along with your patch set.

					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>





More information about the Submit mailing list