Remove archaic register keyword from sys and ANSIfy prototypes while here
ClosedPublic
Actions

Authored by emaste on Mar 30 2017, 12:53 AM.

Details

Reviewers

kib
cem
jhb
sbruno
emaste

Group Reviewers

Intel Networking

Commits

rS331643: MFC r314568 (by emaste):
rS318389: Remove register keyword from sys/ and ANSIfy prototypes

Summary

A long long time ago the register keyword told the compiler to store the corresponding variable in a CPU register, but it is not relevant for any compiler used in the FreeBSD world today.

Diff Detail

Lint

Lint Skipped

Unit

Tests Skipped

Event Timeline

emaste created this revision.Mar 30 2017, 12:53 AM

Herald added a reviewer: Intel Networking. · View Herald TranscriptMar 30 2017, 12:53 AM

Herald added a subscriber: ae. · View Herald Transcript

Did you use a coccinelle script for the ANSIfying, and if so, what was it?

In D10193#210689, @cem wrote:

Did you use a coccinelle script for the ANSIfying, and if so, what was it?

I did not - this is a collection of changes that have been sitting in various WIP trees of mine for ages, through a combination of manual changes and set etc.

emaste added a subscriber: royger.Mar 30 2017, 1:08 AM

cem accepted this revision.Mar 30 2017, 1:09 AM

cem added inline comments.

sys/kern/kern_clock.c
392–394	didn't feel like ANSIfying these too? :)

This revision is now accepted and ready to land.Mar 30 2017, 1:09 AM

emaste added inline comments.Mar 30 2017, 1:15 AM

sys/amd64/include/xen/hypercall.h
112 ↗	(On Diff #26799)	hypercall.h changes will be omitted, the register keyword is required for the `__asm__("r10")`.

emaste added inline comments.Mar 30 2017, 1:18 AM

sys/kern/kern_clock.c
392–394	hrm, missed these. Updated in my tree.

jhb added inline comments.Mar 30 2017, 4:37 PM

sys/cddl/contrib/opensolaris/uts/common/sys/sysmacros.h
391	It would seem worth trying to upstream the sys/cddl/contrib changes or otherwise dropping them.
sys/ddb/db_access.c
59–60	Missed this 'register'.
sys/ddb/db_sym.c
292	Missed this 'register'.
369–370	Missed this 'register'.
sys/kern/sysv_msg.c
768–769	ANSIfy?
sys/kern/vfs_mount.c
368	Maybe follow the more typical '#ifndef _SYSPROTO_H_' format rather than doing this type of custom comment?
sys/kern/vfs_syscalls.c
1549	Use SYS_SYSPROTO_H here as well.
2357	SYS_SYSPROTO_H

Just a LGTM based on the Intel Networking group.

sys/dev/ex/if_ex.c
815 ↗	(On Diff #26799)	Approved.
sys/dev/ixgb/if_ixgb.c
1780	Approved.

Update based on feedback in Phabricator

This revision now requires review to proceed.Apr 6 2017, 5:49 PM

emaste marked 7 inline comments as done.Apr 6 2017, 5:53 PM

emaste added inline comments.

sys/kern/vfs_syscalls.c
1555	leftover will be deleted

emaste marked 2 inline comments as done.May 16 2017, 9:40 PM

emaste added inline comments.

sys/cddl/contrib/opensolaris/uts/common/sys/sysmacros.h
391	Will handle these outside of this review.
sys/dev/ixgb/if_ixgb.c
1780	previously committed

emaste accepted this revision.May 17 2017, 1:53 PM

emaste marked an inline comment as done.

This revision is now accepted and ready to land.May 17 2017, 1:53 PM

emaste closed this revision.May 17 2017, 1:53 PM

A note for posterity: I compared object files built with/without this patch. There were 8 files with differences, and those differences were due to __LINE__ being passed to mtx_lock or similar.