tech-toolchain archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: HEADS UP: I will be merging christos-time_t by the end of the week

On Wed, Jan 07, 2009 at 02:08:54PM +0900, SODA Noriyuki wrote:
> Well, there is a problem in glibc, though.
> It makes system-call-hooking hard.
> Especially because glibc makes many system calls "hidden" (i.e. it
> makes those symbols not use PLT).
> Many checkpointing libraries are suffered by that.

On Linux, the compiler seems to directly generate the syscall entry
sequence for many system calls.  Presumably this is some sort of
'calling convention' defined in the .h file.
(ie the argumets and system call number are put into appropriate registers
and the trap (or whatever) used to enter the kernel.)

Such calls can reside within the program .o file.

In NetBSD there are always assembler system call stubs.


David Laight:

Home | Main Index | Thread Index | Old Index