System call registration mess?
I want to register a new system call and notice that on several architectures there is some inconsistency between the system call table and unistd.h, e.g. (2.6.13-rc2):
in arch/arm/kernel/calls.S:
...
/* 310 */ .long sys_request_key
.long sys_keyctl
.long sys_semtimedop__syscall_end:
...
and in include/asm-arm/unistd.h:
...
#define __NR_request_key (__NR_SYSCALL_BASE+310)
#define __NR_keyctl (__NR_SYSCALL_BASE+311)
#if 0 /* reserved for un-muxing ipc */
#define __NR_semtimedop (__NR_SYSCALL_BASE+312)
#endif
#define __NR_vserver (__NR_SYSCALL_BASE+313)
/*
* The following SWIs are ARM private.
*/
#define __ARM_NR_BASE (__NR_SYSCALL_BASE+0x0f0000)
...
So it seems that sys_vserver is not declared in the system call table. Is there any reason for this inconsistency or is this a bug and should be fixed? If second, should there be something like
.long sys_ni_syscall /* reserved for vserver */
or
.long sys_vserver
in the syscall table?
Similar inconsistencies can be found in other architecture subtrees, e.g. in arm26:
arch/arm26/kernel/calls.S:
...
/* 235 */ .long sys_removexattr
.long sys_lremovexattr
.long sys_fremovexattr
.long sys_tkill__syscall_end:
...
in include/asm-arm26/unistd.h there is a whole bunch of calls that are not registered in the system call table:
#define __NR_tkill (__NR_SYSCALL_BASE+238) #define __NR_sendfile64 (__NR_SYSCALL_BASE+239) #define __NR_futex (__NR_SYSCALL_BASE+240) ... #define __NR_mq_notify (__NR_SYSCALL_BASE+278) #define __NR_mq_getsetattr (__NR_SYSCALL_BASE+279) #define __NR_waitid (__NR_SYSCALL_BASE+280)
Should they all get filled up with sys_ni_syscall definitions or with their corresponding entries so I can enter my own syscall at the bottom with correct numbering or what is the proper way to register a new syscall in these cases?
Thanks,
Wieland