Use the right len.

One more length check.

Fix stuff.

Ping?

Oh wow, it's been a year now. Anyway - I still do want to upstream it :-)

The reason I hadn't changed the other instances is that while trivial, they are somewhat harder to test, and I've been bitten by trivial changes to rarely excercised branches in the past.

Ok, let's put this on hold. More careful benchmarking shows there's something seriously weird wrt performance going on.

Heh, I clearly remember looking at the output.c, which uses OUTBUFSIZ, and thinking "ok, let's follow the convention here". I guess I followed it a bit too closely.

:-D

Ok, fixed the constant name. As for size - it's not any specific benchmark; I'm basically fixing various obvious inefficiencies I see in truss output - it's kind of byproduct of another project. As for the size - my thinking here is that it's pretty much free; we don't keep a number of those buffers around, are we? Given that /etc/rc.subr is 49281, 128k seems like a reasonable choice to me.

Use a better constant name.

Note that the '(!int)' syntax is to avoid spurious differences with the current style there.

"Full Name".

I'd need some help with pushing it upstream - the libutil CMake patch is FreeBSD-specific and I'm not sure how to rewrite it properly.

In D17564#374936, @imp wrote:

The flex man page suggests there's a performance improvement if we use -CF or -Cf as well. -CFr or -Cfr may also be beneficial since it beneficial. The man page warns against increased size, but I doubt the differences that were important when flex was written (1995) are relevant today. If we're looking at performance tuning, we should look here.

Adding -CFr does increase the size, but not be enough to matter:

  9978      4    88   10070   0x2756   nslexer.o.no-CFr
26883      4    80   26967   0x6957   nslexer.o.-CFr

so ~16k increase in size. If you have a way to benchmark things, you might see if this is faster or not.

Fix ordering.

Ping?

Changes suggested by markj@.

As it is now, I believe they could work in FreeBSD. In Linux, glibc already provides the built-in modules as shared libraries, so it would clash there. I've grepped the ports tree for '\<nss_.*so', and I don't see an example of anyone doing that.

Remove unrelated parts of the diff, as suggested by des@.

Having said that, the part in question doesn't belong here. So, I'll redo the diff.

Meh, wrong diff.

Fix issues found by Allan.

Reshuffle the information a bit, so that the syntax description
is in a separate paragraph.

I'm not sure. I think the path is a kind of SDDM-specific override, and the current behaviour makes sense: just update it to the differences between Linux and FreeBSD, like it's done in patch-src_common_Configuration.h right now.

Just for the record, a similar mechanism - although not rtld-specific - seems to have been implemented in Domain/OS, http://www.bitsavers.org/pdf/apollo/014962-A00_Domain_OS_Design_Principles_Jan89.pdf, the "Signal Extensions" chapter.

Improve.