In D59815#1377193, @ehem_freebsd_m5p.com wrote:

What @brooks wrote. Building HEAD on 15.1 doesn't work now and reverting rG:20c09e6fece makes the build work on 15.1.

Decay immediate form to non-immediate when clang is too old

This bit of abstraction would allow a freebsd32_jail_* implementations and will also a 64-bit version on CHERI targets.

Ideally there would be a GCC gate as well, but I couldn't immediately find when support was added.

rG20c09e6fece29dbcf4835b4dd90e969f1b9b3e59 raised the minimum LLVM version to build FreeBSD to 20. That's probably too far given that 15.1 still has 19.

Reading this code again, I did find myself wondering if we should explictly skip the size < maxsize check in the XDR_FREE case instead of overriding maxsize.

A few high-level comments.

In D59380#1363039, @rlibby wrote:

In D59380#1362825, @des wrote:

Even if another program is writing to the same file at the same time, the kernel serializes those writes and each line is written atomically. If the log is on NFS each write is a single request and should be handled atomically by the server, even if it is too long for a single packet.

This is the part that makes me uneasy. This seems to amount to "short writes can't happen". I do not believe write(2) makes any such guarantee and I am familiar with file systems where this is not true. The implementation of the kernel function dofilewrite (sys_write / kern_writev / dofilewrite) seems to support the idea that short writes are a contemplated and supported part of the VFS design. This very program handles short writes in its copy() routine.

Anyway, I agree that an 30% gain is small potatoes when a 15x gain seems possible, so maybe the patch at hand is not the right approach.

Expand struct sendfile_args in kern_sendfile signature

In D59034#1354663, @kib wrote:

Why passing struct sendile_args to kern_sendfile()? This is not how all other kern_* wrappers are done, and IMO is worse than flattening the args. You have to explicitly initialize the literal anyway, which makes it equivalent to list the function's argument in the call.

• freebsd32_copyin_hdtr: take a void *, fix trailer count
• make copyinuio_t take a void * (but don't update the functions yet)

Merging the implementations will avoid future instances of oversights like the one fixed in 4b17776d9afd0009ac8547126c59c97eda0f3fc9 as well as avoiding a third copy of this code from freebsd64 when it lands.

Please name the file following the exiting convention (patch-backport-4bed796a5fea) and produce with --no-prefix.

Seems reasonable to me.

Correct feature introduction to 16.0. Bump doc date.

Put the common case inside the switch statement

Actually describe this test

Improve test description

Move global struct uexterror and constructor to its own file

Switch to rwlock, hoist free outside the locked region.

Fix some whitespace

Move to its own file, document, update copyrights

Make "sys/" prefix depend on category source. Move test to D58237

Move test here from D58238

When an unsupported or broken format is encoutered emite a warning like "<unsupported-format>".

Emit errors in formatted text for unsupported or invalid formats.

Update for UEXTERROR returning void.

Add docs. Make uexterr_set return void.

Use UEXTERROR().

Add UEXTERROR() analog to EXTERROR(). It differs in that it just wraps uexterr_set and thus it gets a different name.