Kernel:
Change statistics to use the *uptime() timescale (ie: relative to
boottime) rather than the UTC aligned timescale. This makes the
device statistics code oblivious to clock steps.
Change timestamps to bintime format, they are cheaper.
Remove the "busy_count", and replace it with two counter fields:
"start_count" and "end_count", which are updated in the down and
up paths respectively. This removes the locking constraint on
devstat.
Add a timestamp argument to devstat_start_transaction(), this will
normally be a timestamp set by the *_bio() function in bp->bio_t0.
Use this field to calculate duration of I/O operations.
Add two timestamp arguments to devstat_end_transaction(), one is
the current time, a NULL pointer means "take timestamp yourself",
the other is the timestamp of when this transaction started (see
above).
Change calculation of busy_time to operate on "the salami principle":
Only when we are idle, which we can determine by the start+end
counts being identical, do we update the "busy_from" field in the
down path. In the up path we accumulate the timeslice in busy_time
and update busy_from.
Change the byte_* and num_* fields into two arrays: bytes[] and
operations[].
Userland:
Change the misleading "busy_time" name to be called "snap_time" and
make the time long double since that is what most users need anyway,
fill it using clock_gettime(CLOCK_MONOTONIC) to put it on the same
timescale as the kernel fields.
Change devstat_compute_etime() to operate on struct bintime.
Remove the version 2 legacy interface: the change to bintime makes
compatibility far too expensive.
Fix a bug in systat's "vm" page where boot relative busy times would
be bogus.
Bump __FreeBSD_version to 500107
Review & Collaboration by: ken
vmstat so that they never coalesce. Both iostat and vmstat need larger
fixes to prevent wide fields from unnecessarily messing up the alignment
of all subsequent fields.
PR: 41674
MFC-after: 3 days
originally written in January, 2000, but have been substantially updated.
- No longer use hz/stathz and the CPU times in computing the TTY stats,
but rather use etime, like the disk stats.
- Clean up malloc/realloc failure tests.
- Use a new integrated routine to fetch devstat information via sysctl or
KVM.
- Get rid of the X() macro for calculating CPU stats
- Use rint() on the CPU state display to avoid truncation errors. (this
requires libm)
- Clean up flag usage somewhat.
Reviewed by: bde
crash dumps, and make it use sysctl for all data retrievals in the
"live" case (i.e. when not using iostat on a crash dump).
Remove setgid kmem for the iostat executable, it is not needed any
more after these changes.
Reviewed by: ken
Remove -? flag that was not working but documented. Make it work instead
but hide it in man page and usage string as others tools do.
Spelling.
Abort on allocation failure (with errx()).
o getopt returns -1 rather than EOF on errors
o getopt returns '?' for characters it doesn't know about, so
don't include them in the getopt options string.
generation was causing unaligned access faults on the Alpha.
I have incremented the devstat version number, since this is an interface
change. You'll need to recompile libdevstat, systat, iostat, vmstat and
rpc.rstatd along with your kernel.
Partially Submitted by: Andrew Gallatin <gallatin@cs.duke.edu>
fixed variable so all manually-specified devices are shown rather then
cutting the display off at 3, fixed formatting for msps when msps is
larger then 100 (it was blowing the columner display output before),
added -K option to make blk count (-oIK) force a 1K block size (to conform
more closely to systat, vmstat definition of blocksize).
