if we don't have an address expression available in a predecessor,
then model this as the value being clobbered at the end of the pred
block instead of being modeled as a complete phi translation failure.
This is important for PRE of loads because we want to see that the
load is available in all but this predecessor, and complete phi
translation failure results in not getting any information about
predecessors.
This doesn't do anything until I renable code insertion since PRE
now sees that it is available in all but one predecessors, but can't
insert the addressing in the predecessor that is missing it to
eliminate the redundancy.
llvm-svn: 90037
translation of add with immediate. This allows us
to optimize this function:
void test(int N, double* G) {
long j;
G[1] = 1;
for (j = 1; j < N - 1; j++)
G[j+1] = G[j] + G[j+1];
}
to only do one load every iteration of the loop.
llvm-svn: 90013
Update all analysis passes and transforms to treat free calls just like FreeInst.
Remove RaiseAllocations and all its tests since FreeInst no longer needs to be raised.
llvm-svn: 84987
so that all code paths get it. PR4256 was about a case where the
phi translation loop would find all preds in the Visited cache, so
it could get by without re-sorting the NonLocalPointerDeps cache.
Fix this by resorting it earlier, there is no reason not to do this.
This patch inspired by Jakub Staszak's patch.
llvm-svn: 75476
This avoids using a dangling pointer.
Reset NumSortedEntries after restoring Cache to avoid extraneous sorts.
This fixes the reduced sqlite3 testcase, but apparently not the whole app.
llvm-svn: 62838
analyses could be run without the caches properly sorted. This
can fix all sorts of weirdness. Many thanks to Bill for coming
up with the 'issorted' verification idea.
llvm-svn: 62757
visited set before they are used. If used, their blocks need to be
added to the visited set so that subsequent queries don't use conflicting
pointer values in the cache result blocks.
llvm-svn: 61080
memdep keeps track of how PHIs affect the pointer in dep queries, which
allows it to eliminate the load in cases like rle-phi-translate.ll, which
basically end up being:
BB1:
X = load P
br BB3
BB2:
Y = load Q
br BB3
BB3:
R = phi [P] [Q]
load R
turning "load R" into a phi of X/Y. In addition to additional exposed
opportunities, this makes memdep safe in many cases that it wasn't before
(which is required for load PRE) and also makes it substantially more
efficient. For example, consider:
bb1: // has many predecessors.
P = some_operator()
load P
In this example, previously memdep would scan all the predecessors of BB1
to see if they had something that would mustalias P. In some cases (e.g.
test/Transforms/GVN/rle-must-alias.ll) it would actually find them and end
up eliminating something. In many other cases though, it would scan and not
find anything useful. MemDep now stops at a block if the pointer is defined
in that block and cannot be phi translated to predecessors. This causes it
to miss the (rare) cases like rle-must-alias.ll, but makes it faster by not
scanning tons of stuff that is unlikely to be useful. For example, this
speeds up GVN as a whole from 3.928s to 2.448s (60%)!. IMO, scalar GVN
should be enhanced to simplify the rle-must-alias pointer base anyway, which
would allow the loads to be eliminated.
In the future, this should be enhanced to phi translate through geps and
bitcasts as well (as indicated by FIXMEs) making memdep even more powerful.
llvm-svn: 61022
of a pointer. This allows is to catch more equivalencies. For example,
the type_lists_compatible_p function used to require two iterations of
the gvn pass (!) to delete its 18 redundant loads because the first pass
would CSE all the addressing computation cruft, which would unblock the
second memdep/gvn passes from recognizing them. This change allows
memdep/gvn to catch all 18 when run just once on the function (as is
typical :) instead of just 3.
On all of 403.gcc, this bumps up the # reundandancies found from:
63 gvn - Number of instructions PRE'd
153991 gvn - Number of instructions deleted
50069 gvn - Number of loads deleted
to:
63 gvn - Number of instructions PRE'd
154137 gvn - Number of instructions deleted
50185 gvn - Number of loads deleted
+120 loads deleted isn't bad.
llvm-svn: 60799
tricks based on readnone/readonly functions.
Teach memdep to look past readonly calls when analyzing
deps for a readonly call. This allows elimination of a
few more calls from 403.gcc:
before:
63 gvn - Number of instructions PRE'd
153986 gvn - Number of instructions deleted
50069 gvn - Number of loads deleted
after:
63 gvn - Number of instructions PRE'd
153991 gvn - Number of instructions deleted
50069 gvn - Number of loads deleted
5 calls isn't much, but this adds plumbing for the next change.
llvm-svn: 60794
load dependence queries. This allows GVN to eliminate a few more
instructions on 403.gcc:
152598 gvn - Number of instructions deleted
49240 gvn - Number of loads deleted
after:
153986 gvn - Number of instructions deleted
50069 gvn - Number of loads deleted
llvm-svn: 60786
the first block of a query specially. This makes the "complete query
caching" subsystem more effective, avoiding predecessor queries. This
speeds up GVN another 4%.
llvm-svn: 60752
track of whether the CachedNonLocalPointerInfo for a block is specific
to a block. If so, just return it without any pred scanning. This is
good for a 6% speedup on GVN (when it uses this lookup method, which
it doesn't right now).
llvm-svn: 60695
method. This will eventually take over load/store dep
queries from getNonLocalDependency. For now it works
fine, but is incredibly slow because it does no caching.
Lets not switch GVN to use it until that is fixed :)
llvm-svn: 60649
clobber with the current implementation. Instead of returning
a "precise clobber" just return a fuzzy one. This doesn't
matter to any clients anyway and should speed up analysis time
very very slightly.
llvm-svn: 60641
1. Merge the 'None' result into 'Normal', making loads
and stores return their dependencies on allocations as Normal.
2. Split the 'Normal' result into 'Clobber' and 'Def' to
distinguish between the cases when memdep knows the value is
produced from when we just know if may be changed.
3. Move some of the logic for determining whether readonly calls
are CSEs into memdep instead of it being in GVN. This still
leaves verification that the arguments are hte same to GVN to
let it know about value equivalences in different contexts.
4. Change memdep's call/call dependency analysis to use
getModRefInfo(CallSite,CallSite) instead of doing something
very weak. This only really matters for things like DSA, but
someday maybe we'll have some other decent context sensitive
analyses :)
5. This reimplements the guts of memdep to handle the new results.
6. This simplifies GVN significantly:
a) readonly call CSE is slightly simpler
b) I eliminated the "getDependencyFrom" chaining for load
elimination and load CSE doesn't have to worry about
volatile (they are always clobbers) anymore.
c) GVN no longer does any 'lastLoad' caching, leaving it to
memdep.
7. The logic in DSE is simplified a bit and sped up. A potentially
unsafe case was eliminated.
llvm-svn: 60607
vector instead of a densemap. This shrinks the memory usage of this thing
substantially (the high water mark) as well as making operations like
scanning it faster. This speeds up memdep slightly, gvn goes from
3.9376 to 3.9118s on 403.gcc
This also splits out the statistics for the cached non-local case to
differentiate between the dirty and clean cached case. Here's the stats
for 403.gcc:
6153 memdep - Number of dirty cached non-local responses
169336 memdep - Number of fully cached non-local responses
162428 memdep - Number of uncached non-local responses
yay for caching :)
llvm-svn: 60313
ReverseLocalDeps when we update it. This fixes a regression test
failure from my last commit.
Second, for each non-local cached information structure, keep a bit that
indicates whether it is dirty or not. This saves us a scan over the whole
thing in the common case when it isn't dirty.
llvm-svn: 60274
instead of containing them by value. This increases the density
(!) of NonLocalDeps as well as making the reallocation case
faster. This speeds up gvn on 403.gcc by 2% and makes room for
future improvements.
I'm not super thrilled with having to explicitly manage the new/delete
of the map, but it is necesary for the next change.
llvm-svn: 60271
If we see that a load depends on the allocation of its memory with no
intervening stores, we now return a 'None' depedency instead of "Normal".
This tweaks GVN to do its optimization with the new result.
llvm-svn: 60267
dependencies. The basic situation was this: consider if we had:
store1
...
store2
...
store3
Where memdep thinks that store3 depends on store2 and store2 depends
on store1. The problem happens when we delete store2: The code in
question was updating dep info for store3 to be store1. This is a
spiffy optimization, but is not safe at all, because aliasing isn't
transitive. This bug isn't exposed today with DSE because DSE will only
zap store2 if it is identifical to store 3, and in this case, it is
safe to update it to depend on store1. However, memcpyopt is not so
fortunate, which is presumably why the "dropInstruction" code used to
exist.
Since this doesn't actually provide a speedup in practice, just rip the
code out.
llvm-svn: 60263
an entry in the nonlocal deps map, don't reset entries
referencing that instruction to [dirty, null], instead, set
them to [dirty,next] where next is the instruction after the
deleted one. Use this information in the non-local deps
code to avoid rescanning entire blocks.
This speeds up GVN slightly by avoiding pointless work. On
403.gcc this makes GVN 1.5% faster.
llvm-svn: 60256
Put a some code back to handle buggy behavior that GVN expects: it wants
loads to depend on each other, and accesses to depend on their allocations.
llvm-svn: 60240
Document the Dirty value more precisely, use it for the uninitialized
DepResultTy value. Change reverse mappings to be from an instruction*
instead of DepResultTy, and stop tracking other forms. This makes it more
clear that we only care about the instruction cases.
Eliminate a DepResultTy,bool pair by using Dirty in the local case as well,
shrinking the map and simplifying the code.
This speeds up GVN by ~3% on 403.gcc.
llvm-svn: 60232
query. This makes it crystal clear what cases can escape from MemDep that
the clients have to handle. This also gives the clients a nice simplified
interface to it that is easy to poke at.
This patch also makes DepResultTy and MemoryDependenceAnalysis::DepType
private, yay.
llvm-svn: 60231
of a pointer/int pair instead of a manually bitmangled pointer.
This forces clients to think a little more about checking the
appropriate pieces and will be useful for internal
implementation improvements later.
I'm not particularly happy with this. After going through this
I don't think that the clients of memdep should be exposed to
the internal type at all. I'll fix this in a subsequent commit.
This has no functionality change.
llvm-svn: 60230
properly updates the reverse dependency map when it installs updated
dependencies for instructions that depend on the removed instruction.
llvm-svn: 60222
circumstances we could end up remapping a dependee to the same instruction
that we're trying to remove. Handle this properly by just falling back to
a conservative solution.
llvm-svn: 54132
into alias analysis. This meant updating the API
which now has versions of the getModRefBehavior,
doesNotAccessMemory and onlyReadsMemory methods
which take a callsite parameter. These should be
used unless the callsite is not known, since in
general they can do a better job than the versions
that take a function. Also, users should no longer
call the version of getModRefBehavior that takes
both a function and a callsite. To reduce the
chance of misuse it is now protected.
llvm-svn: 44487
The meaning of getTypeSize was not clear - clarifying it is important
now that we have x86 long double and arbitrary precision integers.
The issue with long double is that it requires 80 bits, and this is
not a multiple of its alignment. This gives a primitive type for
which getTypeSize differed from getABITypeSize. For arbitrary precision
integers it is even worse: there is the minimum number of bits needed to
hold the type (eg: 36 for an i36), the maximum number of bits that will
be overwriten when storing the type (40 bits for i36) and the ABI size
(i.e. the storage size rounded up to a multiple of the alignment; 64 bits
for i36).
This patch removes getTypeSize (not really - it is still there but
deprecated to allow for a gradual transition). Instead there is:
(1) getTypeSizeInBits - a number of bits that suffices to hold all
values of the type. For a primitive type, this is the minimum number
of bits. For an i36 this is 36 bits. For x86 long double it is 80.
This corresponds to gcc's TYPE_PRECISION.
(2) getTypeStoreSizeInBits - the maximum number of bits that is
written when storing the type (or read when reading it). For an
i36 this is 40 bits, for an x86 long double it is 80 bits. This
is the size alias analysis is interested in (getTypeStoreSize
returns the number of bytes). There doesn't seem to be anything
corresponding to this in gcc.
(3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded
up to a multiple of the alignment. For an i36 this is 64, for an
x86 long double this is 96 or 128 depending on the OS. This is the
spacing between consecutive elements when you form an array out of
this type (getABITypeSize returns the number of bytes). This is
TYPE_SIZE in gcc.
Since successive elements in a SequentialType (arrays, pointers
and vectors) need to be aligned, the spacing between them will be
given by getABITypeSize. This means that the size of an array
is the length times the getABITypeSize. It also means that GEP
computations need to use getABITypeSize when computing offsets.
Furthermore, if an alloca allocates several elements at once then
these too need to be aligned, so the size of the alloca has to be
the number of elements multiplied by getABITypeSize. Logically
speaking this doesn't have to be the case when allocating just
one element, but it is simpler to also use getABITypeSize in this
case. So alloca's and mallocs should use getABITypeSize. Finally,
since gcc's only notion of size is that given by getABITypeSize, if
you want to output assembler etc the same as gcc then getABITypeSize
is the size you want.
Since a store will overwrite no more than getTypeStoreSize bytes,
and a read will read no more than that many bytes, this is the
notion of size appropriate for alias analysis calculations.
In this patch I have corrected all type size uses except some of
those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard
cases). I will get around to auditing these too at some point,
but I could do with some help.
Finally, I made one change which I think wise but others might
consider pointless and suboptimal: in an unpacked struct the
amount of space allocated for a field is now given by the ABI
size rather than getTypeStoreSize. I did this because every
other place that reserves memory for a type (eg: alloca) now
uses getABITypeSize, and I didn't want to make an exception
for unpacked structs, i.e. I did it to make things more uniform.
This only effects structs containing long doubles and arbitrary
precision integers. If someone wants to pack these types more
tightly they can always use a packed struct.
llvm-svn: 43620
alias analysis, adding caching and lazy computation of queries. This will
be used in planned improvements to memory access optimizations.
llvm-svn: 37958
Nick Lewycky