61110aed26
Fix spelling errors.
...
llvm-svn: 102478
2010-04-28 00:36:01 +00:00
4bd85e47bf
further clarify alignment of globals, fix instcombine
...
to not increase the alignment of globals with an assigned
alignment and section.
llvm-svn: 102476
2010-04-28 00:31:12 +00:00
78e00bcb76
improve the global variable alignment description.
...
it is not generally valid for targets to overalign
them when an alignment is specified.
llvm-svn: 102474
2010-04-28 00:13:42 +00:00
b8b85c1ddb
Integrate Jeffery Yasskin's suggestions with respect to
...
traps flowing through memory references, add some text to
better cover phi nodes and externally-visible side effects,
add an example of instructions being control-dependent
on a trap value, and reword some of the existing trap rules.
llvm-svn: 102399
2010-04-26 23:36:52 +00:00
5d284ae8a0
Consolidate the description of volatile operations, now that some of the
...
intrinsics have volatile semantics in addition to the load and store
instructions.
llvm-svn: 102384
2010-04-26 21:21:24 +00:00
48a2588be4
Branching or switching on trap transfers imminent undefined behavior
...
onto control-dependent instructions.
llvm-svn: 102381
2010-04-26 20:54:53 +00:00
b2a709be07
Fix HTML errors that Jeffery Yasskin noticed.
...
llvm-svn: 102376
2010-04-26 20:21:21 +00:00
283a350f4d
PHI nodes also do not necessarily return trap with a trap operand.
...
llvm-svn: 102278
2010-04-24 22:15:58 +00:00
5725580bb6
Switch getelementptr inbounds and sdiv exact from undef to trap.
...
llvm-svn: 102175
2010-04-23 15:23:32 +00:00
ffc9a6b4ac
Add an initial description of a new concept: trap values, and change
...
the definition of the nsw and nuw flags to make use of it.
nsw was introduced to help optimizers answer yes to the following:
// Can we change i from i32 to i64 to eliminate the cast inside the loop?
for (int i = 0; i < n; ++i) A[i] *= 0.1;
// Can we assume that this loop will eventually terminate?
for (int i = 0; i <= n; ++i) A[i] *= 0.1;
In its current form, it isn't truly sufficient for either.
In the first case, if the increment overflows, it'll still have some
valid i32 value; sign-extending it will produce a value which is 33
homogeneous sign bits trailed by 31 independent undef bits. If i is
promoted to i64, it won't have those same values when it reaches that
point. (The compiler could recover here by reasoning about how i is
used by the load, but that's a lot more complicated and isn't always
possible.)
In the second case, there is no value for i which will be greater than
n, so having the increment return undef on overflow doesn't help.
Trap values are a formalization of some existing concepts that we have
about LLVM IR, and give the optimizers a better basis for answering yes
to both questions above.
llvm-svn: 102140
2010-04-22 23:14:21 +00:00
626b02470b
Pointed out by housel on #llvm.
...
llvm-svn: 101395
2010-04-15 20:35:54 +00:00
685db9d59e
typo
...
llvm-svn: 100738
2010-04-08 00:54:34 +00:00
bd4ca62c80
document isvolatile etc.
...
llvm-svn: 100737
2010-04-08 00:53:57 +00:00
508127b5fd
Update langref for memcpy, memmove, and memset
...
llvm-svn: 100611
2010-04-07 06:35:53 +00:00
51065568cd
Have the inst emitter add the !srcloc mdnode to the machine instr.
...
Have the asmprinter use the mdnode to scavenge a source location if
present. Document this nonsense in langref.
llvm-svn: 100607
2010-04-07 05:38:05 +00:00
022a9fb269
various cleanups from daniel
...
llvm-svn: 98528
2010-03-15 04:12:21 +00:00
bbd8bd3257
tidy up and expound more on how half-float works.
...
llvm-svn: 98524
2010-03-14 23:03:31 +00:00
cd4dd9c1db
Document fp16 intrinsics
...
llvm-svn: 98505
2010-03-14 18:42:47 +00:00
a179e4d0a8
add support, testcases, and dox for the new GHC calling
...
convention. Patch by David Terei!
llvm-svn: 98212
2010-03-11 00:22:57 +00:00
5967649780
Add documentation on sibling call optimization. Rename tailcall2.ll test to sibcall.ll.
...
llvm-svn: 97980
2010-03-08 21:05:02 +00:00
9984bd6092
Add metadata example.
...
llvm-svn: 97764
2010-03-04 23:44:48 +00:00
47f2a839ec
attributes are not part of types anymore, patch by James Woodyatt!
...
llvm-svn: 97557
2010-03-02 06:36:51 +00:00
6232f73c7b
PerformTailCallOpt was renamed to GuaranteedTailCallOpt to
...
better reflect its meaning, now that tail call optimizations
are done by default in some cases.
llvm-svn: 97530
2010-03-02 01:08:11 +00:00
a269a0a667
Fix spelling.
...
llvm-svn: 97447
2010-03-01 17:41:39 +00:00
bc4024f72c
Fix HTML.
...
llvm-svn: 97170
2010-02-25 21:23:24 +00:00
1ad1499d14
Fix a typo.
...
llvm-svn: 97144
2010-02-25 16:51:31 +00:00
8848111952
Clarify the description of pointer types, and move the
...
address space content to its own paragraph.
llvm-svn: 97143
2010-02-25 16:50:07 +00:00
8ad777dc75
Update LangRef to match the code; pointers default to being 64-bit.
...
llvm-svn: 96873
2010-02-23 02:44:03 +00:00
9641d06809
Add support for emitting non-temporal stores for DAGs marked
...
non-temporal. Fix from r96241 for botched encoding of MOVNTDQ.
Add documentation for !nontemporal metadata.
Add a simpler movnt testcase.
llvm-svn: 96386
2010-02-16 20:50:18 +00:00
392be58cad
Add support for a union type in LLVM IR. Patch by Talin!
...
llvm-svn: 96011
2010-02-12 20:49:41 +00:00
be5557e86b
Add a new function attribute, 'alignstack'. It will indicate (when the backends
...
implement support for it) that the stack should be forcibly realigned in the
prologue (and the process reversed in the epilogue).
llvm-svn: 95945
2010-02-12 00:31:15 +00:00
74bb06c0f0
Reintroduce the InlineHint function attribute.
...
This time it's for real! I am going to hook this up in the frontends as well.
The inliner has some experimental heuristics for dealing with the inline hint.
When given a -respect-inlinehint option, functions marked with the inline
keyword are given a threshold just above the default for -O3.
We need some experiments to determine if that is the right thing to do.
llvm-svn: 95466
2010-02-06 01:16:28 +00:00
c2f8f16162
move "Metadata Nodes and Metadata Strings" section to the right place in the
...
document and edit it.
llvm-svn: 93559
2010-01-15 21:50:19 +00:00
8444d7536c
Remove the InlineHint attribute. There are no current or planned
...
users.
llvm-svn: 93558
2010-01-15 21:36:30 +00:00
97257f8e40
mention that unwind isn't to be trusted, patch by Dustin Laurence
...
llvm-svn: 93518
2010-01-15 18:08:37 +00:00
d1a8969dd2
Add top level section for named metadata.
...
llvm-svn: 93172
2010-01-11 19:35:55 +00:00
f991bbb0c1
Improve unclear bits and inaccuracies in structure and insertvalue
...
documentation.
Patch by Dustin Laurence!
llvm-svn: 93170
2010-01-11 19:19:26 +00:00
b867746a05
Clarify the requirements for a "tail call" to actually be optimized into a
...
jump. People clearly weren't finding the extra requirements in
CodeGenerator.html.
llvm-svn: 93070
2010-01-09 19:44:16 +00:00
0de4caa57b
improve documentation for linkonce to be less confusing.
...
llvm-svn: 93066
2010-01-09 19:15:14 +00:00
3070e16e6f
Grammar thinko.
...
llvm-svn: 93027
2010-01-08 21:42:39 +00:00
e307348325
NamedMDNode is a collection MDNodes.
...
llvm-svn: 92761
2010-01-05 20:41:31 +00:00
31e39bd3d4
Update docs for bitcode changes. For object size checking we won't
...
work with partial objects so just count the type as a boolean. Update
appropriately.
llvm-svn: 91954
2009-12-23 00:29:49 +00:00
455c5776e2
More updates to objectsize intrinsic docs.
...
llvm-svn: 90644
2009-12-05 02:46:03 +00:00
8aa46bc6fd
Update per Bill's comments. Work in progress.
...
llvm-svn: 90271
2009-12-01 22:28:41 +00:00
7348432ade
First pass at llvm.objectsize documentation.
...
llvm-svn: 90116
2009-11-30 08:03:53 +00:00
31c0e0ee35
Vector types are no longer required to have a power-of-two length.
...
llvm-svn: 90004
2009-11-27 13:38:03 +00:00
ee7425cc56
These code generator limitations have been removed.
...
llvm-svn: 90003
2009-11-27 12:33:22 +00:00
c74bc28297
The inbounds keyword isn't relevant to overindexing of
...
static array types. Thanks to Duncan for pointing this out!
llvm-svn: 86576
2009-11-09 19:01:53 +00:00
66c4d61010
Remove the "special case" for zero-length arrays, and rephrase this
...
paragraph to be more precise.
llvm-svn: 86572
2009-11-09 18:40:39 +00:00
a381effe72
add the ability for TargetData to return information about legal integer
...
datatypes on a given CPU. This is intended to allow instcombine and other
transformations to avoid converting big sequences of operations to an
inconvenient width, and will help clean up after SRoA. See also "Adding
legal integer sizes to TargetData" on Feb 1, 2009 on llvmdev, and PR3451.
Comments welcome.
llvm-svn: 86370
2009-11-07 09:35:34 +00:00
33fef7e4e5
Add missing end-tag.
...
llvm-svn: 85769
2009-11-02 00:25:26 +00:00
7f4a336785
Some formatting changes.
...
llvm-svn: 85768
2009-11-02 00:24:16 +00:00
aa99c94e2a
Revert 85678/85680. The decision is to stay with the current form of
...
indirectbr, thus we don't need "blockaddr(@func, null)". Eliminate it
for simplicity.
llvm-svn: 85699
2009-11-01 01:27:45 +00:00
b2c0e243c5
Make blockaddress(@func, null) be valid, and make 'deleting a basic
...
block with a blockaddress still referring to it' replace the invalid
blockaddress with a new blockaddress(@func, null) instead of a
inttoptr(1).
This changes the bitcode encoding format, and still needs codegen
support (this should produce a non-zero value, referring to the entry
block of the function would also be quite reasonable).
llvm-svn: 85678
2009-10-31 20:08:37 +00:00
03ab4dcff9
use metavariable <result> instead of SSA name %result for consistency
...
llvm-svn: 85388
2009-10-28 13:14:50 +00:00
bd0328f668
ooops, SSA name should not be part of the link
...
llvm-svn: 85387
2009-10-28 13:05:07 +00:00
f50fd57543
advertise new syntax for unnamed instructions
...
and eliminate confusing double-use of SSA names
(work in progress)
llvm-svn: 85385
2009-10-28 09:21:30 +00:00
d04cb6d0fa
rename indbr -> indirectbr to appease the residents of #llvm.
...
llvm-svn: 85351
2009-10-28 00:19:10 +00:00
b4f2a24363
typo
...
llvm-svn: 85322
2009-10-27 21:52:54 +00:00
5c5f0ac362
you can't take the address of the entry block of a function.
...
llvm-svn: 85321
2009-10-27 21:49:40 +00:00
da37b30b21
improvements from gabor.
...
llvm-svn: 85320
2009-10-27 21:44:20 +00:00
2bfd320226
fix things pointed out by Dan!
...
llvm-svn: 85310
2009-10-27 21:19:13 +00:00
e4801f7844
document the forthcoming blockaddress constant.
...
llvm-svn: 85306
2009-10-27 21:01:34 +00:00
6506d6c3d7
fix pasto pointed out by Rafael
...
llvm-svn: 85294
2009-10-27 20:27:24 +00:00
3ed871fe62
add enough support for indirect branch for the feature test to pass
...
(assembler,asmprinter, bc reader+writer) and document it. Codegen
currently aborts on it.
llvm-svn: 85274
2009-10-27 19:13:16 +00:00
d20fd59e29
Fix reversed logic spotted by Owen Anderson.
...
llvm-svn: 85251
2009-10-27 16:56:58 +00:00
a70c6dff97
Remove all references to MallocInst and FreeInst
...
llvm-svn: 85177
2009-10-26 23:44:29 +00:00
1cfb958b00
Rename msasm to alignstack per review.
...
llvm-svn: 84795
2009-10-21 23:28:00 +00:00
63c94fe1e2
Documentation for the new msasm flag, which is no
...
worse than the rest of the asm documentation.
llvm-svn: 84037
2009-10-13 21:56:55 +00:00
9bc890425a
Force memory use markers to have a ConstantInt for the size argument.
...
llvm-svn: 83960
2009-10-13 07:57:33 +00:00
6f7d8347c1
Add new "memory use marker" intrinsics. These indicate lifetimes and invariant
...
sections of memory objects.
llvm-svn: 83953
2009-10-13 07:03:23 +00:00
0f4203861f
Documentation: Perform automated correction of common typos.
...
llvm-svn: 83849
2009-10-12 14:46:08 +00:00
80d73c7ee7
fix broken anchors.
...
llvm-svn: 83715
2009-10-10 18:26:06 +00:00
93e06a57c9
Remove the "metadata*" type and simplify the code it complicated. This was only
...
used to support GlobalVariables storing MDNodes, back when they were derived
from Constant before the introduction of NamedMDNode, but never removed.
llvm-svn: 82943
2009-09-27 23:27:42 +00:00
14d1cccc2b
Remove dead code from this function and optimize. Update its corresponding
...
LangRef entry too.
llvm-svn: 82890
2009-09-27 07:55:32 +00:00
84a1eeb7ed
Move the integer type out of 'derived' and into 'primitive'. This permits us
...
to explain that derived types are all composed of other types, which primitive
types aren't. Without moving integer out of derived, this wouldn't be true.
Perform a few trivial cleanups; 'i1' went from a link to #t_primitive to
#t_integer (a holdover from when it was a bool type I suppose).
llvm-svn: 82884
2009-09-27 00:45:11 +00:00
722212d1a0
Revise C library functions description to be vaguer, per Chris.
...
llvm-svn: 82776
2009-09-25 17:04:42 +00:00
68f971b1d6
Clarify that llvm attaches C language semantics to
...
functions with names that match the C library.
llvm-svn: 82701
2009-09-24 18:38:21 +00:00
92ada5dc52
more typos
...
llvm-svn: 81499
2009-09-11 01:49:31 +00:00
10ff0c1862
another typo
...
llvm-svn: 81243
2009-09-08 19:45:34 +00:00
6760e54c92
fix a couple typos pointed out by edwin and duncan
...
llvm-svn: 81219
2009-09-08 15:13:16 +00:00
a34a718c4b
add some more notes.
...
llvm-svn: 81170
2009-09-07 23:33:52 +00:00
ec72b9bbcf
describe undef semantics in some more detail.
...
llvm-svn: 81167
2009-09-07 22:52:39 +00:00
957b1316e6
Fix the syntax of add/sub/mul nsw/nuw and sdiv exact.
...
llvm-svn: 80805
2009-09-02 17:31:42 +00:00
2aaf539564
Add an 'inline hint' attribute to represent source
...
code hints that it would be a good idea to inline
a function ("inline" keyword). No functional change
yet; FEs do not emit this and inliner does not use it.
llvm-svn: 80063
2009-08-26 01:08:21 +00:00
e7f064ed2b
improve documentation for linker_private linkage.
...
llvm-svn: 79893
2009-08-24 04:32:16 +00:00
2de532c132
Clarify a few things in the inbounds description.
...
Describing inbounds in terms of infinitely precise arithmetic prohibits
the assumption address-space-sized integer arithmetic will wrap, with
a wrapped-around value landing within the same object.
llvm-svn: 79538
2009-08-20 17:08:17 +00:00
0aff0b2b58
common globals may also not be marked constant.
...
llvm-svn: 78169
2009-08-05 05:41:44 +00:00
d055488c72
Clarify common linkage and the requirements on it. Enforce
...
them in the verifier.
llvm-svn: 78160
2009-08-05 05:21:07 +00:00
61acaaa0f0
Add one-past-the-end language to the inbounds keyword.
...
llvm-svn: 77460
2009-07-29 16:00:30 +00:00
a40b912352
some wording changes.
...
llvm-svn: 77443
2009-07-29 06:44:13 +00:00
8ae4e24c67
Fix a few typos and add links.
...
llvm-svn: 77403
2009-07-29 00:02:05 +00:00
1639c3905e
Add a new keyword 'inbounds' for use with getelementptr. See the
...
LangRef.html changes for details.
llvm-svn: 77259
2009-07-27 21:53:46 +00:00
e39f03351d
Fix typo.
...
llvm-svn: 77244
2009-07-27 20:37:10 +00:00
6154a01cee
Following discussion on llvm-dev ("proposed new rule for getelementptr"),
...
add a new "Pointer Aliasing Rules" section.
llvm-svn: 77216
2009-07-27 18:07:55 +00:00
6b8677015d
Update to new syntax.
...
llvm-svn: 77043
2009-07-25 02:23:48 +00:00
902dfff8b6
Rename the new unsigned and signed keywords to nuw and nsw,
...
which stand for no-unsigned-wrap and no-signed-wrap.
llvm-svn: 76810
2009-07-22 22:44:56 +00:00
71dfd782ce
Misc. doc fixes following suggestions from Eli.
...
llvm-svn: 76699
2009-07-22 00:04:19 +00:00
b07de44d5a
Documentation for the new non-overflow and exact keywords.
...
llvm-svn: 76495
2009-07-20 22:41:19 +00:00
58f9bb2ccd
implement a new magic global "llvm.compiler.used" which is like llvm.used, but
...
doesn't cause ".no_dead_strip" to be emitted on darwin.
llvm-svn: 76399
2009-07-20 06:14:25 +00:00
Dan Gohman