Bump sys_errlist size to keep ABI backward-compatible for some time.
ClosedPublic
Actions

Authored by kib on Dec 25 2018, 2:19 PM.

Details

Reviewers

emaste
yuripv
mckusick
jilles

Commits

rS342551: Bump sys_errlist size to keep ABI backward-compatible for some time.

Summary

Addition of the new errno values requires adding new elements to sys_errlist array, which is actually ABI-incompatible, since ELF records the object size. Expand array in advance to 150 elements so that we have to go over the issue only once.

Test Plan

I checked that perror(1) printed expected strings for negative, normal errno, and errnos 96, 97, 149, 150, 151 in C and ru_RU.KOI8-R locales.

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

kib created this revision.Dec 25 2018, 2:19 PM

Herald added a subscriber: imp. · View Herald TranscriptDec 25 2018, 2:19 PM

Harbormaster completed remote builds in B21726: Diff 52313.Dec 25 2018, 2:19 PM

kib edited the test plan for this revision. (Show Details)Dec 25 2018, 2:20 PM

yuripv accepted this revision.Dec 25 2018, 2:57 PM

yuripv added inline comments.

lib/libc/string/strerror.c
55 ↗	(On Diff #52313)	Not directly related to your change, but I don't think this is/was correct -- if we have NLS enabled, and catalog for current language has the 32767 (0xffff) entry corresponding to "unknown error", and that text is longer than we calculate here, and the num is big enough, the resulting message will be truncated. I understand that it's really too much "if"s, just noting that errstr() likely should just allocate the buffer big enough instead; that is (20 + 2 + strlen(uprefix) + 1).

This revision is now accepted and ready to land.Dec 25 2018, 2:57 PM

yuripv added inline comments.Dec 25 2018, 3:02 PM

lib/libc/string/strerror.c
55 ↗	(On Diff #52313)	Err, that comment got me confused, we don't actually copy the uprefix to tmp, right? So we don't need to reserve the space for __uprefix there?

kib added inline comments.Dec 25 2018, 3:25 PM

lib/libc/string/strerror.c
55 ↗	(On Diff #52313)	It seems we don't do that now. It was changed in r108603, 2003.

Update EBUFSIZE definition and comment.

This revision now requires review to proceed.Dec 25 2018, 3:33 PM

Harbormaster completed remote builds in B21728: Diff 52317.Dec 25 2018, 3:33 PM

yuripv added inline comments.Dec 25 2018, 3:39 PM

lib/libc/string/strerror.c
52 ↗	(On Diff #52317)	Sorry for making you do the unrelated change.. I think this should now read based on what we actually write into `tmp`: Define buffer big enough to contain delimiter (": ", 2 bytes), 64-bit signed integer converted to ASCII decimal (19 bytes) with optional leading sign (1 byte), and a trailing NUL. The definition of EBUFSIZE should be 23 then according to the above.

kib added inline comments.Dec 25 2018, 3:57 PM

lib/libc/string/strerror.c
52 ↗	(On Diff #52317)	Well, num is int which is 32-bit value and not 64-bit, on any supported architecture. That said, lets pretend that we would support ILP64 or larger arch one day.