Make linux_errtbl[] more readable

[trasz]

Summary

Add defines for Linux errno values and use them to make linux_errtbl[] more readable. While here, add linux_check_errtbl()
function to make sure we don't leave holes.

[trasz]

Note that I'd prefer to get https://reviews.freebsd.org/D26970 committed first.

[emaste]

I think this is good, I'm not sure the /* [80, 89] */ range markers are really needed

sys/compat/linux/linux_errno.c

31 ↗

(On Diff #78813)

do you want to make this < 0 instead

sys/compat/linux/linux_errno.h

81 ↗

(On Diff #78813)

could we indicate what makes these XXX-worthy?

[lwhsu]

sys/compat/linux/linux_errno.h

5 ↗

(On Diff #78813)

We don't really need this line anymore. :-)

[trasz]

In D26972#601660, @emaste wrote:

I think this is good, I'm not sure the /* [80, 89] */ range markers are really needed

I'd prefer to leave them in; they were quite useful when writing this, and are still useful if you want to look up errno by value.

sys/compat/linux/linux_errno.c

31 ↗

(On Diff #78813)

I'd like to get rid of the minus signs in subsequent commit.

sys/compat/linux/linux_errno.h

5 ↗

(On Diff #78813)

Thanks, I'll remove it :-)

81 ↗

(On Diff #78813)

Sure, something like "/* XXX: in Linux this errno value is not defined */"?

[emaste]

sys/compat/linux/linux_errno.h

81 ↗

(On Diff #78813)

Actually, with that said why do we provide a define for it? What about e.g. /* XXX errno 41 unused in Linux */

sys/compat/linux/linux_errno.inc

119 ↗

(On Diff #78813)

These comments were necessary before because it wasn't clear which FreeBSD and Linux errnos are involved, but now it just repeats what the code clearly shows. I think it is still useful to call out discrepancies, but IMO we ought to instead explain why we chose the mapping.

EPROCLIM Too many processes.

EAGAIN Resource temporarily unavailable. This is a temporary condition

and later calls to the same routine may complete normally.

[trasz]

Regen.

[trasz]

sys/compat/linux/linux_errno.h

81 ↗

(On Diff #78813)

It will be required for subsequent commit which adds Linux->FreeBSD mapping.

[trasz]

Readd linux_errno.inc.

[trasz]

sys/compat/linux/linux_errno.inc

119 ↗

(On Diff #78813)

In this particular case I have no idea why the mapping is like this. I just preserved the comments, and added them to point out discrepancies which were previously undocumented. In other words: I agree we should revisit them, but for now it's just maintaining status quo.

[trasz]

Okay, actually emaste@ is right, drop the fake defines.

[emaste]

sys/compat/linux/linux_errno.h

81 ↗

(On Diff #78813)

Ah, so you are going to make an inverse table using the same defines.

SEEMSUNUSED and ALSOUNUSED seem odd though.

How about LINUX_EUNUSED_41 and LINUX_EUNUSED_58

sys/compat/linux/linux_errno.inc

119 ↗

(On Diff #78813)

Sure, but you can drop the "X -> Y" from the comment, just leave an XXX to highlight it

[trasz]

Drop redundant comments.

[trasz]

sys/compat/linux/linux_errno.h

81 ↗

(On Diff #78813)

I think I'd prefer to just use raw numerical values instead of "fake" Econstants.

sys/compat/linux/linux_errno.inc

119 ↗

(On Diff #78813)

Done. I've added a comment at the top of the file to explain what the 'XXX's mean.

[emaste]

Ok no objections to this version

[trasz]

Try to fix tinderbox.

[emaste]

lib/libsysdecode/errno.c

40 ↗

(On Diff #78868)

the static seems odd there

[trasz]

Make it less odd.

[gallatin]

Isn't errno different in linux on different architectures? I know that at least alpha had a very different errno than x86..

[trasz]

It is; at some point we might need to move this into architecture-specific include. Unless Linux folks managed to fix it for newer architectures, in which case we won’t.