PR lib/56395
PR toolchain/56771

Fix profiling on CPUs that do not support unaligned memory access;
Allocate objects (referenced from struct gmonparam) with proper
alignments.

For monstartup(), objects are allocated on heap. Break is not
guaranteed to be aligned at all, unlike stack pointer.

For _m_gmon_alloc(), objects are allocated on anonymous memory.
p->tos is not aligned properly in general.

This fixes quasi-random crashes for *_profile tests, at least on
SH-4 and PowerPC 403 [1]. Also, no regression is observed for
others as far as I can see.

This change does not cause any ABI breakage, as long as application
uses proper pointers; use p->tos instead of evil pointer arithmetic
like (struct tostruct *)((char *)p->froms + p->fromssize) [2].

[1] Timeout should be increased for some tests. "pic" variants
still fail as expected. Dynamically-linked binaries also crash in
rtld for SH-4, but this seems different problem...

[2] This example did not work even before, since the order of
froms[] and tos[] is reversed depending on which of monstartup() or
_m_gmon_alloc() is used for allocation.

PR/56246: Yuichiro Naito: gprof(1) never show call graphs when profiling
multi-threaded application

make more descriptors that we open as close-on-exec

PR/45989: Martin Husemann: lint invocation does include -w only on i386

- turn lint -w for all the platforms after fixing the lint warnings.
- add _DIAGASSERTS() for casts that would assign values to types that
would not fit.
- change types, add casts
- change into ansii prototypes
- turn on _DIAGNOSTIC for libc (during current, to be eliminated for release
builds)

approved by core@

branches: 1.33.6;
Fix fdleak in #ifdef DEBUG. Found by cppcheck.

- KNF
- fix hertz()
- fix error and warning messages

Close log file handle after using it. Found by cppcheck.

sign-compare fixes

branches: 1.29.28;
fix incomplete initializers

don't include <sys/param.h> twice.

use rownddown/rowndup in sys/param.h instead of their now-deleted UC equivs
formerly in sys/gmon.h.

in _m_gmon_destructor(0 when traversing the list, add (void *) to the cast
chain.

fix debugging printf formats.

fix compilation with -DDEBUG

WARNS=4

Make gmon / mcount work properly for threaded applications by keeping
thread-specific gmonparam structures and then merging them before dumping
the profile buffers.

Written by Nathan J. Williams, contributed by Wasabi Systems, Inc.

branches: 1.21.2;
Move UCB-licensed code from 4-clause to 3-clause licence.

Patches provided by Joel Baker in PR 22280, verified by myself.

Use snprintf instead of hand crafted string copy and numeric conversion.
Avoids problems if/when the pid is greater than 999999.
(approved by christos)

Fix signed/unsigned comparison warnings.

disable profiling set{u,g}id binaries.

branches: 1.17.2;
convert from __progname to getprogname()

fix nested extern
fix sbrk casts

branches: 1.15.10;
* On architectures using ELF, and conditionally on those in transition, rename
_ASM_LABEL(cerror) and _ASM_LABEL(curbrk) to _C_LABEL(__cerror) and
_C_LABEL(__curbrk) (or their respective architecture-specific equivalents) to
avoid possible name clashes with identifiers used in user applications.
* Do the same for minbrk on all architectures to avoid a GCC-specific (and
on ELF architectures effectively useless) symbol reference renaming in MI code.

delint

bzero->memset, bcopy->memcpy, bcmp->memcmp

Prefer STDERR_FILENO over 2.

remove obsolete register declarations

Fix RCSID's
Use warn instead of perror()
Use "namespace.h"
Add missing prototypes

be safe with buffers. KNF.

- convert unsafe strcpy(), strcat() and sprintf() to the `n' versions.
- some KNF.

use __inline and __asm, rather than inline and asm. By default (without -g)
lint won't accept the latter two, but will accept the former two as valid.
As far as gcc's concerned, they're the same.

branches: 1.6.2;
use p->hashfraction when doing non-time-critical calculations, rather than
using HASHFRACTION directly. in time-critical calculations, if HASHFRACTION
is a power of two, check that p->hashfraction == HASHFRACTION and if so do
the calculation with the compiled-in value so that the compiler can optimize
out (potentially) expensive divisions. if p->hashfraction != HASHFRACTION,
actually do the division. This has the result that on machines with slow
division, the division can be optimized out of the common case, but that
if HASHFRACTION changes from the compiled-in value (for whatever reason),
profiling will still work. Changes suggested by Chris Torek.

branches: 1.5.4;
Add support for PROFDIR environment variable

kill a gratuitous cast of lowpc to an int

clean up Ids

branches: 1.2.4;
rcsids. also, avoid floating point when picking scale value. doesn't
cost much, only done once.

branches: 1.1.1;
Initial revision