re: misc/58515 (struct acct at acct(5) man page is different from /usr/include/sys/acct.h)

["matthew green via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR misc/58515; it has been noted by GNATS.

From: matthew green <mrg%eterna23.net@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: christos%NetBSD.org@localhost, misc-bug-people%netbsd.org@localhost,
    netbsd-bugs%netbsd.org@localhost, gnats-admin%netbsd.org@localhost, riastradh%NetBSD.org@localhost,
    hashikaw%mail.ru@localhost
Subject: re: misc/58515 (struct acct at acct(5) man page is different from /usr/include/sys/acct.h)
Date: Wed, 09 Oct 2024 15:51:40 +1100

 riastradh%NetBSD.org@localhost writes:
 > Synopsis: struct acct at acct(5) man page is different from /usr/include=
 /sys/acct.h
 >
 > State-Changed-From-To: open->feedback
 > State-Changed-By: riastradh%NetBSD.org@localhost
 > State-Changed-When: Tue, 08 Oct 2024 23:29:15 +0000
 > State-Changed-Why:
 > Fixed in HEAD?  Needs pullups?
 
 i've been meaning to say... acct(5) was affected in netbsd 6.x
 and beyond, as the size of time_t *and* dev_t members grew, but
 there is no compat code etc:
 
 struct acct {
 ...
         time_t    ac_btime;     /* starting time */
 ...
         dev_t     ac_tty;       /* controlling tty */
 
 
 if we're gonna consider fixing this, i'd consider doing it as
 part of an 3rd change that bumped the sizes of various things
 that are kinda limited in 2024 (vs 1984):
 
 /*
  * Accounting structures; these use a comp_t type which is a 3 bits base 8
  * exponent, 13 bit fraction ``floating point'' number.  Units are 1/AHZ
  * seconds.
  */
 typedef uint16_t comp_t;
 ...
         comp_t    ac_utime;     /* user time */
         comp_t    ac_stime;     /* system time */
         comp_t    ac_etime;     /* elapsed time */
 ...
         comp_t    ac_io;        /* count of IO blocks */
 
 limited in both ways:
 
 - *tiny* cpu time is far more possible now
 - *massive threaded* cpu time is far more possible now.
 
 
 .mrg.