But there is also dma, sendmail and parts of kerberos… :-)

Wouldn't /usr/libexec be a better place for it, like for many other such daemons (like fingerd etc…)?

sort it alphabetically, actually

rewrite the return path

don't return too early

Style and s/Int/Sizet/ where appropriate.

Address @kib comments.

Add space after comma.

Use constants defined in D21693.

add a missing file

Address @kib's comments.

Update to r351542 and bump __FreeBSD_version again.

Add a lightweight compat layer in libc.

Check for ENOSYS.

Split the function for one that do the actual work and teach sys___sysctlbyname and freebsd32___sysctlbyname to call it properly.

Update diff to r351302 and address some comments.

I confirm that it works.

This breaks ifconfig_DEFAULT="DHCP" somehow. It works with either SYNCDHCP or this commit reverted.

Update diff with minor fixes.

Man: start a sentence from new line.

Add CAVEATS section to the man page to point out the rounding issue with too short buffers.

In D13578#284349, @kib wrote:

In one branch of the snprintf() use, we add divisor / 2 and remainder * 10, not remainder. Does this case need to be handled differently ?

In D13578#283989, @cem wrote:

I don't understand the actual algorithmic change, but given the strong test coverage, I think I'm ok with this. Some style nits below.

Address @cem comments about style(9).

I'm fine with it. I've one question about what is the correct way of sorting local variables? style(9) use two different versions (or I'm reading it wrong):

int
main(int argc, char *argv[])
{
        char *ep;
        long num;
        int ch;

Address @kib comments. Also more and more style(9) cleanup like moving variables from body to top of the functions, removing unneeded empty lines, PRIu64 changed to %ju with uintmax_t cast, sorting, etc...

vmstat(1) is full of style(9) violations, mostly related to libxo integration. I'd like to clean this in a separate patch.

Move uintmax_t cast to vmstat_pgtok macro.

Change u_int64_t to uint64_t in prthuman() and all invocations.

s/divider/divisor/

Rename real_putuint64() to do_putuint64(), style(9) fixes.

No need to use 64 bit arithmetics for namei cache.

Use correctly generated diff.

Split printing integer values to putint() and putuint64() functions, calling the actual implementation with parameter selecting using SI or IEC divisor. Update all memory related values to use putuint64().

In D13105#273075, @kib wrote:

In D13105#272897, @pawel.biernacki-gmail.com wrote:

Address @kib comments:

• Drop 1000 as a divider and stick to expected 1024.

So we have two uses of putint(). One is for (possibly very large) VM counters, and others for (usually not too large) event counts e.g. syscall/intr and so on.
Should the new function used only for 'large' counters, leaving old display algorithm around for slow counters ? This question jibes with the note that changed function should be called putuint64.

Address @kib comments:

• Update with long diff.
• Use correct formatting for uint64_t in snprintf.

Sorry, uploaded the short patch. Will upload long one with fixes.

Fix the alignment issue by introducing spare filed of 3x16 bit ints.

Reordering didn't really helped. There is 5 16-bit fields that seems to be aligned on amd64 leaving 3x16bit hole which is only 1x16bit on i386.
Any ideas how to fix it? Enforce struct alignment to 8 bytes? Add an extra 32bit field as a spare?

Address the style(9) comments. Also there is no need to call sysctl_handle_opaque in the last return as all we do now is returning fully populated struct. SYSCTL_OUT is enough.

Address @kib comments:

• Extend all counters to uint64_t.
• Introduce new P_OSREL_VMTOTAL64 define to mark extension of the struct vmtotal.
• Provide sysctl(8) compatibility.
• Bump FreeBSD_version.

From what I see, sysctl(8) first calls sysctl(3) to get the size of vmtotal. As defined in

SYSCTL_PROC(_vm, VM_TOTAL, vmtotal, CTLTYPE_OPAQUE|CTLFLAG_RD|CTLFLAG_MPSAFE,
    0, sizeof(struct vmtotal), vmtotal, "S,vmtotal", 
    "System virtual memory statistics");

it'll return the size of the current vmtotal. It later compares it with the compile time size of struct vmtotal and report mismatch.

Address @mjg comments and remove extra FILEDESC locks.