[AArch64] Return the correct size for TLSDESC_CALLSEQ
The branch relaxation pass is computing the wrong offsets because it assumes
TLSDESC_CALLSEQ eats up 4 bytes, when in fact it is lowered to an instruction
sequence taking up 16 bytes. This can become a problem in huge files with lots
of TLS accesses, as it may slowly move branch targets out of the range computed
by the branch relaxation pass.
Fixes PR24234 https://llvm.org/bugs/show_bug.cgi?id=24234
Differential Revision: https://reviews.llvm.org/D22870
This fixes "error in backend: fixup value out of range" when compiling
the misc/talkfilters port for AArch64.
Reported by: sbruno
PR: 201762
MFC after: 3 days
Fix for pr24346: arm asm label calculation error in sub
Some ARM instructions encode 32-bit immediates as a 8-bit integer
(0-255) and a 4-bit rotation (0-30, even) in its least significant 12
bits. The original fixup, FK_Data_4, patches the instruction by the
value bit-to-bit, regardless of the encoding. For example, assuming
the label L1 and L2 are 0x0 and 0x104 respectively, the following
instruction:
add r0, r0, #(L2 - L1) ; expects 0x104, i.e., 260
would be assembled to the following, which adds 1 to r0, instead of
260:
e2800104 add r0, r0, #4, 2 ; equivalently 1
The new fixup kind fixup_arm_mod_imm takes care of the encoding:
e2800f41 add r0, r0, #260
Patch by Ting-Yuan Huang!
This fixes label calculation for ARM assembly, and is needed to enable
ARM assembly sources for OpenSSL.
Requested by: jkim
MFC after: 3 days
[X86] AMD Bobcat CPU (btver1) doesn't support XSAVE
btver1 is a SSSE3/SSE4a only CPU - it doesn't have AVX and doesn't
support XSAVE.
Differential Revision: http://reviews.llvm.org/D17682
Pull in r262782 from upstream llvm trunk (by Simon Pilgrim):
[X86] AMD Bobcat CPU (btver1) doesn't support XSAVE
btver1 is a SSSE3/SSE4a only CPU - it doesn't have AVX and doesn't
support XSAVE.
Differential Revision: http://reviews.llvm.org/D17683
This ensures clang does not emit AVX instructions for CPUTYPE=btver1.
Reported by: Michel Depeige <demik+freebsd@lostwave.net>
PR: 211864
MFC after: 3 days
Only attempt to detect AVG if SSE2 is available
Summary:
In PR29973 Sanjay Patel reported an assertion failure when a certain
loop was optimized, for a target without SSE2 support. It turned out
this was because of the AVG pattern detection introduced in rL253952.
Prevent the assertion failure by bailing out early in
`detectAVGPattern()`, if the target does not support SSE2.
Also add a minimized test case.
Reviewers: congh, eli.friedman, spatel
Subscribers: emaste, llvm-commits
Differential Revision: http://reviews.llvm.org/D20905
This should fix assertion failures ("Requires at least SSE2!") when
building the games/0ad port with CPUTYPE=pentium3.
Reported by: madpilot
[VectorUtils] Fix nasty use-after-free
In truncateToMinimalBitwidths() we were RAUW'ing an instruction then
erasing it. However, that intruction could be cached in the map we're
iterating over. The first check is "I->use_empty()" which in most
cases would return true, as the (deleted) object was RAUW'd first so
would have zero use count. However in some cases the object could
have been polluted or written over and this wouldn't be the case.
Also it makes valgrind, asan and traditionalists who don't like their
compiler to crash sad.
No testcase as there are no externally visible symptoms apart from a
crash if the stars align.
Fixes PR26509.
This should fix crashes when building a number of ports on arm64.
Reported by: andrew
[X86] Emit a proper ADJCALLSTACKDOWN in EmitLoweredTLSAddr
We forgot to add the second machine operand to our ADJCALLSTACKDOWN,
resulting in crashes in PEI.
This fixes PR27071.
This should fix an assertion failure during buildworld, when using -Os,
and targeting either i386 directly, or building the 32-bit libraries on
amd64.
Reported by: Eric Camachat <eric.camachat@gmail.com>
the safe point to insert the prologue and epilogue of the function) on
X86. This prevents problems with some functions using TLS, such as in
jemalloc, and which was the cause for Address Sanitizer crashes. The
correct fix is still being discussed upstream.
[DwarfDebug] Move MergeValues to .cpp, NFC
Pull in r257979 from upstream llvm trunk, by Keno Fischer:
[DwarfDebug] Don't merge DebugLocEntries if their pieces overlap
Summary:
Later in DWARF emission we check that DebugLocEntries have
non-overlapping pieces, so we should create any such entries
by merging here.
Fixes PR26163.
Reviewers: aprantl
Differential Revision: http://reviews.llvm.org/D16249
Again, these will be merged to the official release_38 branch soon, but
we need them ASAP.
be merged to the official release_38 branch soon, but we need it ASAP):
Stop increasing alignment of externally-visible globals on ELF
platforms.
With ELF, the alignment of a global variable in a shared library will
get copied into an executables linked against it, if the executable even
accesss the variable. So, it's not possible to implicitly increase
alignment based on access patterns, or you'll break existing binaries.
This happened to affect libc++'s std::cout symbol, for example. See
thread: http://thread.gmane.org/gmane.comp.compilers.clang.devel/45311
(This is a re-commit of r257719, without the bug reported in
PR26144. I've tweaked the code to not assert-fail in
enforceKnownAlignment when computeKnownBits doesn't recurse far enough
to find the underlying Alloca/GlobalObject value.)
Differential Revision: http://reviews.llvm.org/D16145
from upstream clang trunk, which sets the default debug tuning back to
gdb. The lldb debug tuning is not yet grokked completely by our ELF
manipulation tools.
bugfix-only release, with no new features.
Please note that from 3.5.0 onwards, clang and llvm require C++11
support to build; see UPDATING for more information.
[x86] Fix wrong lowering of vsetcc nodes (PR25080).
Function LowerVSETCC (in X86ISelLowering.cpp) worked under the wrong
assumption that for non-AVX512 targets, the source type and destination type
of a type-legalized setcc node were always the same type.
This assumption was unfortunately incorrect; the type legalizer is not always
able to promote the return type of a setcc to the same type as the first
operand of a setcc.
In the case of a vsetcc node, the legalizer firstly checks if the first input
operand has a legal type. If so, then it promotes the return type of the vsetcc
to that same type. Otherwise, the return type is promoted to the 'next legal
type', which, for vectors of MVT::i1 is always a 128-bit integer vector type.
Example (-mattr=+avx):
%0 = trunc <8 x i32> %a to <8 x i23>
%1 = icmp eq <8 x i23> %0, zeroinitializer
The initial selection dag for the code above is:
v8i1 = setcc t5, t7, seteq:ch
t5: v8i23 = truncate t2
t2: v8i32,ch = CopyFromReg t0, Register:v8i32 %vreg1
t7: v8i32 = build_vector of all zeroes.
The type legalizer would firstly check if 't5' has a legal type. If so, then it
would reuse that same type to promote the return type of the setcc node.
Unfortunately 't5' is of illegal type v8i23, and therefore it cannot be used to
promote the return type of the setcc node. Consequently, the setcc return type
is promoted to v8i16. Later on, 't5' is promoted to v8i32 thus leading to the
following dag node:
v8i16 = setcc t32, t25, seteq:ch
where t32 and t25 are now values of type v8i32.
Before this patch, function LowerVSETCC would have wrongly expanded the setcc
to a single X86ISD::PCMPEQ. Surprisingly, ISel was still able to match an
instruction. In our case, ISel would have matched a VPCMPEQWrr:
t37: v8i16 = X86ISD::VPCMPEQWrr t36, t25
However, t36 and t25 are both VR256, while the result type is instead of class
VR128. This inconsistency ended up causing the insertion of COPY instructions
like this:
%vreg7<def> = COPY %vreg3; VR128:%vreg7 VR256:%vreg3
Which is an invalid full copy (not a sub register copy).
Eventually, the backend would have hit an UNREACHABLE "Cannot emit physreg copy
instruction" in the attempt to expand the malformed pseudo COPY instructions.
This patch fixes the problem adding the missing logic in LowerVSETCC to handle
the corner case of a setcc with 128-bit return type and 256-bit operand type.
This problem was originally reported by Dimitry as PR25080. It has been latent
for a very long time. I have added the minimal reproducible from that bugzilla
as test setcc-lowering.ll.
Differential Revision: http://reviews.llvm.org/D13660
This should fix the "Cannot emit physreg copy instruction" errors when
compiling contrib/wpa/src/common/ieee802_11_common.c, and CPUTYPE is set
to a CPU supporting AVX (e.g. sandybridge, ivybridge).
[SLP] Vectorize for all-constant entries.
This should fix libc++'s iostream initialization SIGBUSing on amd64,
whenever the global cout symbol is not aligned to 16 bytes.
Some further explanation: libc++'s iostream.cpp contains the definitions
of std::cout, std::cerr and so on. These global objects are effectively
declared with an alignment of 8 bytes. When an executable is linked
against libc++.so, it can sometimes get a copy of the global object,
which is then at the same alignment.
However, with clang 3.7.0, the initialization of these global objects
will incorrectly use SSE instructions (e.g. movdqa), whenever the
optimization level is high enough, and SSE is enabled, such as on amd64.
When any of these objects is not aligned to 16 bytes, this will result
in a SIGBUS during iostream initialization. In contrast, clang 3.6.x
and earlier took the 8 byte alignment into consideration, and avoided
SSE for those particular operations.
After bisecting of upstream changes, I found that the above revision
caused the change of this behavior, so I am reverting it now as a
workaround, while a discussion and test case is being prepared for
upstream.
[SCCP] Turn loads of null into undef instead of zero initialized values
Surprisingly, this is a correctness issue: the mmx type exists for
calling convention purposes, LLVM doesn't have a zero representation for
them.
This partially fixes PR23999.
Pull in r241143 from upstream llvm trunk (by David Majnemer):
[LoopUnroll] Use undef for phis with no value live
We would create a phi node with a zero initialized operand instead of
undef in the case where no value was originally available. This was
problematic for x86_mmx which has no null value.
These fix a "Cannot create a null constant of that type!" error when
compiling the graphics/sdl2_gfx port with MMX enabled.
Reported by: amdmi3
MC: Allow multiple comma-separated expressions on the .uleb128 directive.
For compatiblity with GNU as. Binutils documents this as
'.uleb128 expressions'. Subtle, isn't it?
Reported by: sbruno
PR: 199554
MFC after: 3 days
As is described at http://llvm.org/bugs/show_bug.cgi?id=22408, the GNU
linkers ld.bfd and ld.gold currently only support a subset of the
whole range of AArch64 ELF TLS relocations. Furthermore, they assume
that some of the code sequences to access thread-local variables are
produced in a very specific sequence. When the sequence is not as the
linker expects, it can silently mis-relaxe/mis-optimize the
instructions.
Even if that wouldn't be the case, it's good to produce the exact
sequence, as that ensures that linkers can perform optimizing
relaxations.
This patch:
* implements support for 16MiB TLS area size instead of 4GiB TLS area
size. Ideally clang would grow an -mtls-size option to allow support
for both, but that's not part of this patch.
* by default doesn't produce local dynamic access patterns, as even
modern ld.bfd and ld.gold linkers do not support the associated
relocations. An option (-aarch64-elf-ldtls-generation) is added to
enable generation of local dynamic code sequence, but is off by
default.
* makes sure that the exact expected code sequence for local dynamic
and general dynamic accesses is produced, by making use of a new
pseudo instruction. The patch also removes two
(AArch64ISD::TLSDESC_BLR, AArch64ISD::TLSDESC_CALL) pre-existing
AArch64-specific pseudo SDNode instructions that are superseded by
the new one (TLSDESC_CALLSEQ).
Submitted by: Kristof Beyls
Differential Revision: https://reviews.freebsd.org/D2175
ARM: treat [N x i32] and [N x i64] as AAPCS composite types
The logic is almost there already, with our special homogeneous
aggregate handling. Tweaking it like this allows front-ends to emit
AAPCS compliant code without ever having to count registers or add
discarded padding arguments.
Only arrays of i32 and i64 are needed to model AAPCS rules, but I
decided to apply the logic to all integer arrays for more consistency.
This fixes a possible "Unexpected member type for HA" error when
compiling lib/msun/bsdsrc/b_tgamma.c for armv6.
Reported by: Jakub Palider <jpa@semihalf.com>
LoopRotate: When reconstructing loop simplify form don't split edges
from indirectbrs.
Yet another chapter in the endless story. While this looks like we
leave the loop in a non-canonical state this replicates the logic in
LoopSimplify so it doesn't diverge from the canonical form in any way.
http://llvm.org/PR21968
This fixes a "Cannot split critical edge from IndirectBrInst" assertion
failure when building the devel/radare2 port.
PR: 195480, 196987
MFC after: 3 days
[mips] Enable arithmetic and binary operations for the i128 data type.
Summary:
This patch adds support for some operations that were missing from
128-bit integer types (add/sub/mul/sdiv/udiv... etc.). With these
changes we can support the __int128_t and __uint128_t data types
from C/C++.
Depends on D7125
Reviewers: dsanders
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7143
This fixes "error in backend" messages, when compiling parts of
compiler-rt using 128-bit integer types for mips64.
Reported by: sbruno
PR: 197259
[FastIsel][X86] Fix invalid register replacement for bool args
Summary:
Consider the following IR:
%3 = load i8* undef
%4 = trunc i8 %3 to i1
%5 = call %jl_value_t.0* @foo(..., i1 %4, ...)
ret %jl_value_t.0* %5
Bools (that are the result of direct truncs) are lowered as whatever
the argument to the trunc was and a "and 1", causing the part of the
MBB responsible for this argument to look something like this:
%vreg8<def,tied1> = AND8ri %vreg7<kill,tied0>, 1, %EFLAGS<imp-def>; GR8:%vreg8,%vreg7
Later, when the load is lowered, it will insert
%vreg15<def> = MOV8rm %vreg14, 1, %noreg, 0, %noreg; mem:LD1[undef] GR8:%vreg15 GR64:%vreg14
but remember to (at the end of isel) replace vreg7 by vreg15. Now for
the bug. In fast isel lowering, we mistakenly mark vreg8 as the result
of the load instead of the trunc. This adds a fixup to have
vreg8 replaced by whatever the result of the load is as well, so
we end up with
%vreg15<def,tied1> = AND8ri %vreg15<kill,tied0>, 1, %EFLAGS<imp-def>; GR8:%vreg15
which is an SSA violation and causes problems later down the road.
This fixes PR21557.
Test Plan: Test test case from PR21557 is added to the test suite.
Reviewers: ributzka
Reviewed By: ributzka
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6245
This fixes a possible assertion failure when compiling toolbox.cxx from
LibreOffice 4.3.5.
Reported by: kwm
[X86] Convert esp-relative movs of function arguments to pushes, step 2
This moves the transformation introduced in r223757 into a separate MI pass.
This allows it to cover many more cases (not only cases where there must be a
reserved call frame), and perform rudimentary call folding. It still doesn't
have a heuristic, so it is enabled only for optsize/minsize, with stack
alignment <= 8, where it ought to be a fairly clear win.
(Re-commit of r227728)
Differential Revision: http://reviews.llvm.org/D6789
This helps to get sys/boot/i386/boot2 below the required size again,
when optimizing with -Oz.
