| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These two commits:
v5.21.3-759-gff2a62e "Skip no-common-vars optimisation for aliases"
v5.21.4-210-gc997e36 "Make list assignment respect foreach aliasing"
added a run-time mechanism to detect aliased package variables,
by either "*pkg = ...," or "for $pkg (...)", and used that information
to enable the OPpASSIGN_COMMON mechanism at runtime for detecting common
elements in a list assign, e.g.
for $alias ($a, ...) {
($a,$b) = (1,$alias);
}
The previous commit but one changed the OPpASSIGN_COMMON mechanism such
that it no longer uses PL_sawalias. So this var and the mechanism for
setting it can now be removed.
This commit removes:
* the PL_sawalias variable
* the GPf_ALIASED_SV GP flag
* the SAVEt_GP_ALIASED_SV and save_aliased_sv() save type.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
The use of this variable was inconsistent. It was not dup'ed on thread
cloning unless LC_CTYPE is being used, but elsewhere it was. This led
to segfaults on threaded builds. Now it isn't touched anywhere unless
LC_CTYPE is used.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
A previous commit changed how \X is implemented, and now we don't need
these anymore.
|
| |
|
|
|
|
|
|
|
|
|
| |
A function implements seeing if the space between any two characters is
a grapheme cluster break. Afer I wrote this, I realized that an array
lookup might be a better implementation, but the deadline for v5.22 was
too close to change it. I did see that my gcc optimized it down to
an array lookup.
This makes the implementation of \X go from being complicated to
trivial.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 8c6180a91de91a1194f427fc639694f43a903a78 added a warning message
for when Perl determines that the program's underlying locale just
switched into is poorly supported. At the time it was thought that this
would be an extremely rare occurrence. However, a bug in HP-UX -
B.11.00/64 causes this message to be raised for the "C" locale. A
workaround was done that silenced those. However, before it got fixed,
this message would occur gobs of times executing the test suite. It was
raised even if the script is not locale-aware, so that the underlying
locale was completely irrelevant. There is a good prospect that someone
using an older Asian locale as their default would get this message
inappropriately, even if they don't use locales, or switch to a
supported one before using them.
This commit causes the message to be raised only if it actually is
relevant. When not in the scope of 'use locale', the message is stored,
not raised. Upon the first locale-dependent operation within a bad
locale, the saved message is raised, and the storage cleared. I was
able to do this without adding extra branching to the main-line
non-locale execution code. This was done by adding regnodes which get
jumped to by switch statements, and refactoring some existing C tests so
they exclude non-locale right off the bat.
These changes would have been necessary for another locale warning that
I previously agreed to implement, and which is coming a few commits from
now.
I do not know of any way to add tests in the test suite for this. It is
in fact rare for modern locales to have these issues. The way I tested
this was to temporarily change the C code so that all locales are viewed
as defective, and manually note that the warnings came out where
expected, and only where expected.
I chose not to try to output this warning on any POSIX functions called.
I believe that all that are affected are deprecated or scheduled to be
deprecated anyway. And POSIX is closer to the hardware of the machine.
For convenience, I also don't output the message for some zero-length
pattern matches. If something is going to be matched, the message will
likely very soon be raised anyway.
|
| |
|
|
|
| |
Add 5.21.8 in Module::CoreList
Point Maintainers.pl to new version of Module::CoreList
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This op is an optimisation for any series of one or more array or hash
lookups and dereferences, where the key/index is a simple constant or
package/lexical variable. If the first-level lookup is of a simple
array/hash variable or scalar ref, then that is included in the op too.
So all of the following are replaced with a single op:
$h{foo}
$a[$i]
$a[5][$k][$i]
$r->{$k}
local $a[0][$i]
exists $a[$i]{$k}
delete $h{foo}
while these aren't:
$a[0] already handled by OP_AELEMFAST
$a[$x+1] not a simple index
and these are partially replaced:
(expr)->[0]{$k} the bit following (expr) is replaced
$h{foo}[$x+1][0] the first and third lookups are each done with
a multideref op, while the $x+1 expression and
middle lookup are done by existing add, aelem etc
ops.
Up until now, aggregate dereferencing has been very heavyweight in ops; for
example, $r->[0]{$x} is compiled as:
gv[*r] s
rv2sv sKM/DREFAV,1
rv2av[t2] sKR/1
const[IV 0] s
aelem sKM/DREFHV,2
rv2hv sKR/1
gvsv[*x] s
helem vK/2
When executing this, in addition to the actual calls to av_fetch() and
hv_fetch(), there is a lot of overhead of pushing SVs on and off the
stack, and calling lots of little pp() functions from the runops loop
(each with its potential indirect branch miss).
The multideref op avoids that by running all the code in a loop in a
switch statement. It makes use of the new UNOP_AUX type to hold an array
of
typedef union {
PADOFFSET pad_offset;
SV *sv;
IV iv;
UV uv;
} UNOP_AUX_item;
In something like $a[7][$i]{foo}, the GVs or pad offsets for @a and $i are
stored as items in the array, along with a pointer to a const SV holding
'foo', and the UV 7 is stored directly. Along with this, some UVs are used
to store a sequence of actions (several actions are squeezed into a single
UV).
Then the main body of pp_multideref is a big while loop round a switch,
which reads actions and values from the AUX array. The two big branches in
the switch are ones that are affectively unrolled (/DREFAV, rv2av, aelem)
and (/DREFHV, rv2hv, helem) triplets. The other branches are various entry
points that handle retrieving the different types of initial value; for
example 'my %h; $h{foo}' needs to get %h from the pad, while '(expr)->{foo}'
needs to pop expr off the stack.
Note that there is a slight complication with /DEREF; in the example above
of $r->[0]{$x}, the aelem op is actually
aelem sKM/DREFHV,2
which means that the aelem, after having retrieved a (possibly undef)
value from the array, is responsible for autovivifying it into a hash,
ready for the next op. Similarly, the rv2sv that retrieves $r from the
typeglob is responsible for autovivifying it into an AV. This action
of doing the next op's work for it complicates matters somewhat. Within
pp_multideref, the autovivification action is instead included as the
first step of the current action.
In terms of benchmarking with Porting/bench.pl, a simple lexical
$a[$i][$j] shows a reduction of approx 40% in numbers of instructions
executed, while $r->[0][0][0] uses 54% fewer. The speed-up for hash
accesses is relatively more modest, since the actual hash lookup (i.e.
hv_fetch()) is more expensive than an array lookup. A lexical $h{foo}
uses 10% fewer, while $r->{foo}{bar}{baz} uses 34% fewer instructions.
Overall,
bench.pl --tests='/expr::(array|hash)/' ...
gives:
PRE POST
------ ------
Ir 100.00 145.00
Dr 100.00 165.30
Dw 100.00 175.74
COND 100.00 132.02
IND 100.00 171.11
COND_m 100.00 127.65
IND_m 100.00 203.90
with cache misses unchanged at 100%.
In general, the more lookups done, the bigger the proportionate saving.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 8771da69db30134352181c38401c7e50753a7ee8.
Pad lists need to carry IDs around with them, so that when something
tries to close over a pad, it is possible to confirm that the right
pad is being closed over (either the original outer pad, or a clone of
it). (See the commit message of db4cf31d1, in which commit I added an
ID to the padlist struct.)
In 8771da69 I found that I could use the memory address of the pad’s
name list (name lists are shared) and avoid the extra field.
Some time after 8771da69 I realised that a pad list could be freed,
and the same address reused for another pad list, so using a memory
address may not be so wise. I thought it highly unlikely, though, and
put it on the back burner.
I have just run into that. t/comp/form_scope.t is now failing
for me with test 13, added by db4cf31d1. It bisects to 3d6de2cd1
(PERL_PADNAME_MINIMAL), but that’s a red herring. Trivial changes
to the script make the problem go away. And it only happens on non-
debugging builds, and only on my machine. Stepping through with gdb
shows that the format-cloning is following the format prototype’s out-
side pointer and confirming that it is has the correct pad (yes, the
memory addresses are the same), which I know it doesn’t, because I can
see what the test is doing.
While generation numbers can still fall afoul of the same problem, it
is much less likely.
Anyway, the worst thing about 8771da69 is the typo in the first word
of the commit message.
|
| |
|
|
|
| |
These will replace the current use of &PL_sv_undef and &PL_sv_no as
pad names.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The encoding pragma is deprecated, but in the meantime it causes spooky
action at a distance with other modules that it may be combined with.
In these modules, operations such as chr(), ord(), and utf8::upgrade()
will suddenly start doing the wrong thing.
The documentation for 'encoding' has said to call it after loading other
modules, but this may be impractical. This is especially bad with
anything that auto-loads at first use, like \N{} does now for charnames.
There is an issue with combining this with setting the variable
${^ENCODING} directly. The potential for conflicts has always been
there, and remains. This commit introduces a shadow hidden variable,
subservient to ${^ENCODING} (to preserve backwards compatibility) that
has lexical scope validity.
The pod for 'encoding' has been revamped to be more concise, clear, use
more idiomatic English, and to speak from a modern perspective.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- this improves the error message on ABI incompatibility, per
[perl #123136]
- reduce the number of gv_fetchfile calls in newXS over registering many
XSUBs
- "v" was not stripped from PERL_API_VERSION_STRING since string
"vX.XX.X\0", a typical version number is 8 bytes long, and aligned to
4/8 by most compilers in an image. A double digit maint release is
extremely unlikely.
- newXS_deffile saves on machine code in bootstrap functions by not passing
arg filename
- move newXS to where the rest of the newXS*()s live
- move the "no address" panic closer to the start to get it out of the way
sooner flow wise (it nothing to do with var gv or cv)
- move CvANON_on to not check var name twice
- change die message to use %p, more efficient on 32 ptr/64 IV platforms
see ML post "about commit "util.c: fix comiler warnings""
- vars cv/xs_spp (stack pointer pointer)/xs_interp exist for inspection by
a C debugger in an unoptimized build
|
| |
|
|
|
| |
Commit 0e42d607f5 made PL_apiversion unused. Remove it to save memory in
interp struct.
|
| | |
|
| |
|
|
|
|
|
|
|
| |
This prevents perl recursing infinitely when an overloaded object is
assigned to $DB::single, $DB::trace or $DB::signal
This is done by referencing their values as IVs instead of as SVs in
dbstate, and by adding magic to those variables so that assignments to
the scalars update the PL_DBcontrol array.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The ‘no common vars’ optimisation allows perl to copy the values
straight from the rhs to the lhs in a list assignment.
In ($a,$b) = ($c,$d), that means $c gets assigned to $a,
then $d to $b.
If the same variable occurs on both sides of the expression
(($a,$b)=($b,$a)), then it is necessary to make temporary copies of
the variables on the rhs, before assigning them to the left.
If some variables have been aliased to others, then the common vars
detection can be fooled:
*x = *y;
$x = 3;
($x, $z) = (1, $y);
That assigns 1 to $x, and then goes to assign $y to $z, but $y is
the same as $x, which has just been clobbered. So 1 gets assigned
instead of 3.
This commit solves this by recording in each typeglob whether the sca-
lar is an alias of a scalar from elsewhere.
If such a glob is encountered, then the entire expression is ‘tainted’
such that list assignments will assume there might be common vars.
|
| |
|
|
|
|
|
|
| |
This commit allows Perl to be compiled with a bitmap size that is larger
than 256. This bitmap is used to directly look up whether a character
matches or not, without having to do a binary search or hash lookup. It
might improve the performance for some installations that have a lot of
use of scripts that are above the Latin1 range.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
PL_padix keeps track of the position in the pad when pad_alloc has to
start scanning for an available slot.
The availability of a slot is determined differently for targets
(which may reuse slots that are already targets from previous state-
ments, at least when pad_reset is enabled) and constants (which may
not reuse targets).
Having the same index for both may require scanning the entire pad for
allocating a constant or GV.
t/re/uniprops.t was running far too slowly under USE_BROKEN_PAD_RESET
because of this. pad_reset would reset PL_padix to point to the
beginning of a pad with a few hundred thousand entries. pad_alloc
would then have to scan the entire pad before adding a GV to the end.
It is still too slow, even with this commit, but for other reasons.
(This is just a partial fix.)
|
| |
|
|
|
|
|
| |
This description has been a little faulty for quite some time, not
just because of the last few commits. Any time pad_free or pad_swipe
is called, that pad slot becomes available and PL_padix is reset.
So ‘max used’ is not accurate.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
MAD = Misc Attribute Decoration; unmaintained attempt at preserving
the Perl parse tree more faithfully so that automatic conversion to
Perl 6 would have been easier.
|
| |
|
|
|
|
|
|
|
|
|
| |
When processing version strings, the radix character must be a dot even
if we otherwise would be using some other character. vutil.c
upg_version() changes to the dot, but calls sv_catpvf() which may try to
change the character to something else. This commit introduces a way to
lock the character to a dot around the call to sv_catpvf()
vutil.c is cpan-upstream, but already blead and cpan have diverged, so
this just updates the SHA of the new version
|
| |
|
|
| |
This was scheduled to be removed in 5.20, but was forgotten.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
In certain places in the documentation, "5.20" is no longer applicable.
Also, a message referred to in perldiag got reworded, but our checks did
not catch that perldiag should have been updated.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
PL_timesbuf is effectively a vestige of Perl 1, and doesn't actually need to
be an interpreter variable. It will be removed early in v5.21.x, but it's a
good idea to refactor the code not to use it before then. A local struct tms
will be on the C stack, which will be in the CPU's L1 cache, whereas the
relevant part of the interpreter struct may well not be in the CPU cache at
all. Therefore this change might reduce cache pressure fractionally. A local
variable access should also be simpler machine code on most CPU architectures.
|
| |
|
|
| |
on the grounds that its a reasonably hot variable.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This large (sorry, I couldn't figure out how to meaningfully split it
up) commit causes Perl to fully support LC_CTYPE operations (case
changing, character classification) in UTF-8 locales.
As a side effect it resolves [perl #56820].
The basics are easy, but there were a lot of details, and one
troublesome edge case discussed below.
What essentially happens is that when the locale is changed to a UTF-8
one, a global variable is set TRUE (FALSE when changed to a non-UTF-8
locale). Within the scope of 'use locale', this variable is checked,
and if TRUE, the code that Perl uses for non-locale behavior is used
instead of the code for locale behavior. Since Perl's internal
representation is UTF-8, we get UTF-8 behavior for a UTF-8 locale.
More work had to be done for regular expressions. There are three
cases.
1) The character classes \w, [[:punct:]] needed no extra work, as
the changes fall out from the base work.
2) Strings that are to be matched case-insensitively. These form
EXACTFL regops (nodes). Notice that if such a string contains only
characters above-Latin1 that match only themselves, that the node can be
downgraded to an EXACT-only node, which presents better optimization
possibilities, as we now have a fixed string known at compile time to be
required to be in the target string to match. Similarly if all
characters in the string match only other above-Latin1 characters
case-insensitively, the node can be downgraded to a regular EXACTFU node
(match, folding, using Unicode, not locale, rules). The code changes
for this could be done without accepting UTF-8 locales fully, but there
were edge cases which needed to be handled differently if I stopped
there, so I continued on.
In an EXACTFL node, all such characters are now folded at compile time
(just as before this commit), while the other characters whose folds are
locale-dependent are left unfolded. This means that they have to be
folded at execution time based on the locale in effect at the moment.
Again, this isn't a change from before. The difference is that now some
of the folds that need to be done at execution time (in regexec) are
potentially multi-char. Some of the code in regexec was trivial to
extend to account for this because of existing infrastructure, but the
part dealing with regex quantifiers, had to have more work.
Also the code that joins EXACTish nodes together had to be expanded to
account for the possibility of multi-character folds within locale
handling. This was fairly easy, because it already has infrastructure
to handle these under somewhat different circumstances.
3) In bracketed character classes, represented by ANYOF nodes, a new
inversion list was created giving the characters that should be matched
by this node when the runtime locale is UTF-8. The list is ignored
except under that circumstance. To do this, I created a new ANYOF type
which has an extra SV for the inversion list.
The edge case that caused the most difficulty is folding involving the
MICRO SIGN, U+00B5. It folds to the GREEK SMALL LETTER MU, as does the
GREEK CAPITAL LETTER MU. The MICRO SIGN is the only 0-255 range
character that folds to outside that range. The issue is that it
doesn't naturally fall out that it will match the CAP MU. If we let the
CAP MU fold to the samll mu at compile time (which it can because both
are above-Latin1 and so the fold is the same no matter what locale is in
effect), it could appear that the regnode can be downgraded away from
EXACTFL to EXACTFU, but doing so would cause the MICRO SIGN to not case
insensitvely match the CAP MU. This could be special cased in regcomp
and regexec, but I wanted to avoid that. Instead the mktables tables
are set up to include the CAP MU as a character whose presence forbids
the downgrading, so the special casing is in mktables, and not in the C
code.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
This global array is no longer used, having been removed in previous
commits in this series.
Since it is a global, consideration need be given to possible uses of it
outside the core. It has never been externally documented, and is an
opaque structure whose internals have changed with every release. The
functions used to access it are almost all static to regcomp.c; those
few that aren't have been hidden from all but the few .c files that need
to have access to them, via #if's.
|
