build: note that MAKEOBJDIRPREFIX must be absolute
ClosedPublic
Actions

Authored by emaste on Nov 13 2024, 3:15 PM.

Details

Reviewers

imp

Commits

rG75af1320f4b5: build: note that MAKEOBJDIRPREFIX must be absolute

Summary

PR: 282711

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Not Applicable

Unit

Tests Not Applicable

Event Timeline

@sjg should we make some note in make(1) also?

emaste added a reviewer: imp.Nov 13 2024, 3:23 PM

Note: We're a bit inconsistent about referring to MAKEOBJDIRPREFIX as a make variable and an environment variable in the rest of the doc...

share/man/man7/build.7
515	"It" likely is fine here or maybe "This variable" like the prior sentence. But this is also fine if you prefer to make it explicit.

This revision is now accepted and ready to land.Nov 13 2024, 4:16 PM

Closed by commit rG75af1320f4b5: build: note that MAKEOBJDIRPREFIX must be absolute (authored by emaste). · Explain WhyNov 13 2024, 5:45 PM

This revision was automatically updated to reflect the committed changes.

emaste added a commit: rG75af1320f4b5: build: note that MAKEOBJDIRPREFIX must be absolute.

In D47538#1084342, @emaste wrote:

@sjg should we make some note in make(1) also?

Wouldn't hurt, will update in NetBSD first.

Revision Contents
Changeset List

Path

Size

share/

man/

man7/

build.7

4 lines

Diff 146361

View Options

share/man/man7/build.7

Show All 40 Lines
*		*
* CRC32 code derived from work by Gary S. Brown.		* CRC32 code derived from work by Gary S. Brown.
*/		*/

#include <sys/cdefs.h>		#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");		__FBSDID("$FreeBSD$");

#include <sys/param.h>		#include <sys/param.h>
		#include <sys/libkern.h>
#include <sys/systm.h>		#include <sys/systm.h>

		#if defined(__amd64__) \|\| defined(__i386__)
		#include <machine/md_var.h>
		#include <machine/specialreg.h>
		#endif

const uint32_t crc32_tab[] = {		const uint32_t crc32_tab[] = {
0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, 0x706af48f,		0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, 0x706af48f,
0xe963a535, 0x9e6495a3, 0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988,		0xe963a535, 0x9e6495a3, 0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988,
0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, 0x90bf1d91, 0x1db71064, 0x6ab020f2,		0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, 0x90bf1d91, 0x1db71064, 0x6ab020f2,
0xf3b97148, 0x84be41de, 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7,		0xf3b97148, 0x84be41de, 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7,
0x136c9856, 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9,		0x136c9856, 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9,
0xfa0f3d63, 0x8d080df5, 0x3b6e20c8, 0x4c69105e, 0xd56041e4, 0xa2677172,		0xfa0f3d63, 0x8d080df5, 0x3b6e20c8, 0x4c69105e, 0xd56041e4, 0xa2677172,
0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b, 0x35b5a8fa, 0x42b2986c,		0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b, 0x35b5a8fa, 0x42b2986c,
▲ Show 20 Lines • Show All 685 Lines • ▼ Show 20 Lines	multitable_crc32c(uint32_t crc32c,
return (crc32c_sb8_64_bit(crc32c, buffer, length, to_even_word));		return (crc32c_sb8_64_bit(crc32c, buffer, length, to_even_word));
}		}

uint32_t		uint32_t
calculate_crc32c(uint32_t crc32c,		calculate_crc32c(uint32_t crc32c,
const unsigned char *buffer,		const unsigned char *buffer,
unsigned int length)		unsigned int length)
{		{
if (length < 4) {		if (length < 4) {
		jhbUnsubmitted Not Done Inline Actions You can just use `#ifdef __SSE4_2__`. -msse42 (or -msse4) will define that via the compiler driver. That is what controls 'popcnt' (in the case of popcnt we use it if you set CPUTYPE in your make.conf to specify a CPU that includes popcnt). jhb: You can just use `#ifdef __SSE4_2__`. -msse42 (or -msse4) will define that via the compiler…
		cemAuthorUnsubmitted Not Done Inline Actions Is `-msse4` used to compile this file on i386/amd64? I thought it was off by default, and only enabled on a few files. My build logs actually show `-mno-sse -mno-avx`. cem: Is `-msse4` used to compile this file on i386/amd64? I thought it was off by default, and only…
		markjUnsubmitted Not Done Inline Actions The entries in sys/conf/files.* explicitly specify -msse4. markj: The entries in sys/conf/files.* explicitly specify -msse4.
		cemAuthorUnsubmitted Not Done Inline Actions Only for libkern/x86/crc32_sse42.c. Not libkern/crc32.c (this file). cem: Only for libkern/x86/crc32_sse42.c. Not libkern/crc32.c (this file).
return (singletable_crc32c(crc32c, buffer, length));		return (singletable_crc32c(crc32c, buffer, length));
		#if defined(__amd64__) \|\| defined(__i386__)
		kibUnsubmitted Not Done Inline Actions Do we need defined(_KERNEL) there as well ? Why PC98 is excluded (it seems to be on the way out) ? kib: Do we need defined(_KERNEL) there as well ? Why PC98 is excluded (it seems to be on the way…
		cemAuthorUnsubmitted Not Done Inline Actions Is this code built outside of _KERNEL? tinderbox succeeds, so I don't believe this code is used outside of the kernel. PC98 is excluded because I believe it predates SSE4.2, and I preferred to avoid including dead SSE4.2 code in the PC98 kernel if it would never be used. cem: Is this code built outside of _KERNEL? tinderbox succeeds, so I don't believe this code is…
		} else if ((cpu_feature2 & CPUID2_SSE42) != 0) {
		return (sse42_crc32c(crc32c, buffer, length));
		#endif
} else {		} else {
return (multitable_crc32c(crc32c, buffer, length));		return (multitable_crc32c(crc32c, buffer, length));
}		}
}		}

build: note that MAKEOBJDIRPREFIX must be absoluteClosedPublicActions