Unbloat build config

- 10.5% reduction of o//depend dependency graph
- 8.8% reduction in latency of make command
- Fix issue with temporary file cleanup

There's a new -w option in compile.com that turns off the recent
Landlock output path workaround for "good commands" which do not
unlink() the output file like GNU tooling does.

Our new GNU Make unveil sandboxing appears to have zero overhead
in the grand scheme of things. Full builds are pretty fast since
the only thing that's actually slowed us down is probably libcxx

    make -j16 MODE=rel
    RL: took 85,732,063µs wall time
    RL: ballooned to 323,612kb in size
    RL: needed 828,560,521µs cpu (11% kernel)
    RL: caused 39,080,670 page faults (99% memcpy)
    RL: 350,073 context switches (72% consensual)
    RL: performed 0 reads and 11,494,960 write i/o operations

pledge() and unveil() no longer consider ENOSYS to be an error.
These functions have also been added to Python's cosmo module.

This change also removes some WIN32 APIs and System Five magnums
which we're not using and it's doubtful anyone else would be too
This commit is contained in:
Justine Tunney 2022-08-10 01:32:17 -07:00
parent 133c693650
commit ae5d06dc53
1423 changed files with 2213 additions and 5560 deletions

View file

@ -23,7 +23,6 @@
#include "libc/dce.h"
#include "libc/nexgen32e/msr.h"
#include "libc/nexgen32e/x86feature.h"
#include "libc/runtime/interruptiblecall.h"
#include "libc/runtime/pc.internal.h"
#include "libc/sysv/consts/sig.h"
#include "libc/sysv/errfuns.h"
@ -165,36 +164,12 @@ static privileged dontinline int arch_prctl_openbsd(int code, int64_t addr) {
}
static char g_fsgs_once;
static struct InterruptibleCall g_fsgs_icall;
/**
* Don't bother.
*/
int arch_prctl(int code, int64_t addr) {
void *fn = arch_prctl_fsgsbase;
#if 0
if (!g_fsgs_once) {
g_fsgs_once = true;
if (X86_HAVE(FSGSBASE)) {
g_fsgs_icall.sig = SIGILL;
if (interruptiblecall(&g_fsgs_icall, fn, code, addr, 0, 0) != -1 &&
g_fsgs_icall.returnval != -1) {
/* ivybridge+ (2012) lets us change segment registers without
needing a 700ns system call. cpuid and /proc/cpuinfo will both
report it's available; unfortunately, operating systems have an
added ability to restrict this feature in %cr4, which we're not
even allowed to read lool */
g_fsgs_once = 2;
return 0;
}
}
}
if (g_fsgs_once == 2) {
return arch_prctl_fsgsbase(code, addr);
}
#endif
switch (__hostos) {
case METAL:
return arch_prctl_msr(code, addr);