This works by adding a call to __x86.get_pc_thunk.cx immediately before any
instruction that accesses global data and then assembling the instruction to
use the appropriate offset from CX instead of the absolute address. Some forms
cannot be assembled that way and are rewritten to load the address into CX
first.
-buildmode=pie works now, but is not yet tested.
Fixes#13201 (I think)
Change-Id: I32a8561e7fc9dd4ca6ae3b0e57ad78a6c50bf1f5
Reviewed-on: https://go-review.googlesource.com/17014
Reviewed-by: Ian Lance Taylor <iant@golang.org>
I was prodded into doing this in review comments for the ARM version, and it's
going to make shared libs for 386 easier.
Change-Id: Id12de801b1425b8c6b5736fe91b418fc123a4e40
Reviewed-on: https://go-review.googlesource.com/17012
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Android linker does not handle TLS for us. We set up the TLS slot
for g, as darwin/386,amd64 handle instead. This is disgusting and
fragile. We will eventually fix this ugly hack by taking advantage
of the recent TLS IE model implementation. (Instead of referencing
an GOT entry, make the code sequence look into the TLS variable that
holds the offset.)
The TLS slot for g in android/amd64 assumes a fixed offset from %fs.
See runtime/cgo/gcc_android_amd64.c for details.
For golang/go#10743
Change-Id: I1a3fc207946c665515f79026a56ea19134ede2dd
Reviewed-on: https://go-review.googlesource.com/15991
Reviewed-by: David Crawshaw <crawshaw@golang.org>
It's particularly nice to get rid of the android special cases in the linker.
Change-Id: I516363af7ce8a6b2f196fe49cb8887ac787a6dad
Reviewed-on: https://go-review.googlesource.com/14197
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
liblink was rewriting xor by a negative zero (used by SSA
for negation) as XORPS reg,reg.
Fixes strconv.
Change-Id: I627a0a7366618e6b07ba8f0ad0db0e102340c5e3
Reviewed-on: https://go-review.googlesource.com/14200
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
This seems to have broken arm64 in a mysterious way. Will try again later.
This reverts commit 0a3c991fd3.
Change-Id: Ic1b53413c4168977a27381d9cc6fb8d9d7cbb780
Reviewed-on: https://go-review.googlesource.com/12245
Reviewed-by: Russ Cox <rsc@golang.org>
When the prologue call to morestack was moved down to the
bottom of the function, the pc/sp tables were not updated.
If a traceback through a call to morestack is needed, it would
get confused at and stop at morestack.
Confirmed the fix by adding //go:systemstack (which calls
morestackc, but same issue) where it did not belong
and inspecting the crash.
Change-Id: Id0294bb9dba51ef1a49154637228fb57f1086a94
Reviewed-on: https://go-review.googlesource.com/12144
Reviewed-by: Rob Pike <r@golang.org>
It is almost never set and Addr is large, so having the full struct
in the Prog wastes memory most of the time.
Before (on a 64-bit system):
$ sizeof -p cmd/internal/obj Addr Prog
Addr 80
Prog 376
$
After:
$ sizeof -p cmd/internal/obj Addr Prog
Addr 80
Prog 304
$
Change-Id: I491f201241f87543964a7d0f48b85830759be9d0
Reviewed-on: https://go-review.googlesource.com/10457
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
It shrinks Prog type from 448 bytes down to 376 bytes on amd64.
It also makes sense, because I don't know of any modern architecture
that have instructions which can write to two destinations, none of
which is a register (even x86 doesn't have such instructions).
Change-Id: I3061f1c9ac93d79ee2b92ecb9049641d0e0f6300
Reviewed-on: https://go-review.googlesource.com/10330
Reviewed-by: Aram Hăvărneanu <aram@mgk.ro>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Russ Cox <rsc@golang.org>
When reading the object files for linking, liblink takes care of
generate the data for them.
This is a port of https://golang.org/cl/3101 to Go.
Change-Id: Ie3e2d6515bd7d253a8c1e25c70ef8fed064436d8
Signed-off-by: Shenghou Ma <minux@golang.org>
Reviewed-on: https://go-review.googlesource.com/8383
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This started out as trying to remove Bool2int calls, which it does a bit, but
mostly it ended up being removing the Link.Symmorestack array which seemed a
pointless bit of caching.
Change-Id: I91a51eb08cb4b08f3f9f093b575306499267b67a
Reviewed-on: https://go-review.googlesource.com/9239
Reviewed-by: Minux Ma <minux@golang.org>
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The long comment block in obj6.go:progedit talked about the two code sequences
for accessing g as "local exec" and "initial exec", but really they are both forms
of local exec. This stuff is confusing enough without using the wrong words for
things, so rewrite it to talk about 2-instruction and 1-instruction sequences.
Unfortunately the confusion has made it into code, with the R_TLS_IE relocation
now doing double duty as meaning actual initial exec when externally linking and
boring old local exec when linking internally (half of this is my fault). So this
stops using R_TLS_IE in the local exec case. There is a chance this might break
plan9 or windows, but I don't think so. Next step is working out what the heck is
going on on ARM...
Change-Id: I09da4388210cf49dbc99fd25f5172bbe517cee57
Reviewed-on: https://go-review.googlesource.com/9273
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
This lets us avoid loading string constants via the GOT and (together with
http://golang.org/cl/9102) results in the fannkuch benchmark having very similar
register usage with -dynlink as without.
Change-Id: Ic3892b399074982b76773c3e547cfbba5dabb6f9
Reviewed-on: https://go-review.googlesource.com/9103
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
There were 10 implementations of the trivial bool2int function, 9 of which
were the only thing in their file. Remove all of them in favor of one in
cmd/internal/obj.
Change-Id: I9c51d30716239df51186860b9842a5e9b27264d3
Reviewed-on: https://go-review.googlesource.com/9230
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Jumping to an offset past a symbol isn't something that is really
supported by dynamic linkers, so do it by hand.
Change-Id: Ifff8834c6cdfa3d521ebd8479d2e93906df9b258
Reviewed-on: https://go-review.googlesource.com/8238
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
A quick pass through link.go, mostly removing fields that are not
used on the "creating a single object file" side of the fence.
Change-Id: I35ba41378c2c418f7df2f2f88dce65bc64a1a45d
Reviewed-on: https://go-review.googlesource.com/7672
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The c2go translation left a lot of case expressions on separate lines.
Merge expressions onto single lines subject to these constraints:
* Max 4 clauses, all literals or names
* Don't move expressions with comments
The change was created by running http://play.golang.org/p/yHajs72h-g:
$ mergecase cmd/internal/{ld,gc,obj}/*.go cmd/internal/obj/*/*.go
Passes toolstash -cmp.
Change-Id: Iba41b390d302e5486e5dc6ba7599a92270676556
Reviewed-on: https://go-review.googlesource.com/7593
Reviewed-by: Russ Cox <rsc@golang.org>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
This CL moves the bulk of the code that has been copy-and-pasted
since the initial 386 port back into a shared place, cutting 5 copies to 1.
The motivation here is not cleanup per se but instead to reduce the
cost of introducing changes in shared concepts like regalloc or general
expression evaluation. For example, a change after this one will
implement x.(*T) without a call into the runtime. This CL makes that
followup work 5x easier.
The single copy still has more special cases for architecture details
than I'd like, but having them called out explicitly like this at least
opens the door to generalizing the conditions and smoothing out
the distinctions in the future.
This is a LARGE CL. I started by trying to pull in one function at a time
in a sequence of CLs and it became clear that everything was so
interrelated that it had to be moved as a whole. Apologies for the size.
It is not clear how many more releases this code will matter for;
eventually it will be replaced by Keith's SSA work. But as noted above,
the deduplication was necessary to reduce the cost of working on
the current code while we have it.
Passes tests on amd64, 386, arm, and ppc64le.
Can build arm64 binaries but not tested there.
Being able to build binaries means it is probably very close.
Change-Id: I735977f04c0614f80215fb12966dfe9bbd1f5861
Reviewed-on: https://go-review.googlesource.com/7853
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
An interface{} is more in the spirit of the original union.
By my calculations, on 64-bit systems this reduces
Addr from 120 to 80 bytes, and Prog from 592 to 424 bytes.
Change-Id: I0d7b0981513c2a3c94c9ac76bb4f8816485b5a3c
Reviewed-on: https://go-review.googlesource.com/7744
Reviewed-by: Rob Pike <r@golang.org>
These were introduced during C -> Go translation when the loop increment
contained multiple statements.
Change-Id: Ic8abd8dcb3308851a1f7024de00711f0f984e684
Reviewed-on: https://go-review.googlesource.com/7627
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Rob Pike <r@golang.org>
some x86 instructions (e.g. PINSRW) might store memory address in Prog.From3,
so we must also rewrite Prog.From3 on nacl.
Change-Id: I2a0da0f692ba321eba17fbc454d68aaafa133515
Signed-off-by: Shenghou Ma <minux@golang.org>
Reviewed-on: https://go-review.googlesource.com/7074
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>
Remove the per-achitecture formatter for Prog and replace it with
a global String method. Clean up and regularize the output. Update
tests affected by the format; some tests are made correct now when
they were broken before (and known to be).
Also, related: Change the encoding of the (R1+R2) syntax on ppc64
to be equivalent to (R1)(R2*1), which means it needs no special
handling.
Delete the now unused STRINGSZ constant.
Change-Id: I7f6654d11f80065f3914a3f19353f2f12edfe310
Reviewed-on: https://go-review.googlesource.com/6931
Reviewed-by: Russ Cox <rsc@golang.org>
(Because that's what the assembly files actually say - no $ on the constant.)
Change-Id: Idb774cdca0e089c4ac24ab665e23290bf7b565bf
Reviewed-on: https://go-review.googlesource.com/6895
Reviewed-by: Rob Pike <r@golang.org>
Make cmd/internal/obj/x86 support 32-bit mode and use
instead of cmd/internal/obj/i386. Delete cmd/internal/obj/i386.
Clean up encoding of PINSRQ, CMPSD to use explicit third arg
instead of jamming it into an unused slot of a different arg.
Also fix bug in old6a, which declared the wrong grammar.
The accepted (and encoded) arguments to CMPSD etc are mem,reg not reg,mem.
Code that did try to use mem,reg before would be rejected by liblink,
so only reg,reg ever worked, so existing code is not affected.
After this change, code can use mem,reg successfully.
The real bug here is that the encoding tables inverted the argument
order, making the comparisons all backward from what they say on the page.
It's too late to swap them, though: people have already written code that
expects the inverted comparisons (like in package math, and likely externally).
The best we can do is make the argument that should and can take a
memory operand accept it.
Bit-for-bit compatibility checked against tree without this CL.
Change-Id: Ife5685bc98c95001f64407f35066b34b4dae11c1
Reviewed-on: https://go-review.googlesource.com/6810
Reviewed-by: Rob Pike <r@golang.org>
The creation of liblink and subsequent introduction of more explicit
TLS handling broke 6l's (unsupported) -shared flag. This change adds
-shared flags to cmd/asm and 6g and changes liblink to generate shared-
library compatible instruction sequences when they are passed, and
changes 6l to emit the appropriate ELF relocation.
A proper fix probably also requires go tool changes.
Fixes#9652.
Change-Id: I7b7718fe7305c802ac994f4a5c8de68cfbe6c76b
Reviewed-on: https://go-review.googlesource.com/4321
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Run rsc.io/grind rev 796d0f2 on C->Go conversions.
This replaces various awkward := initializations with plain var declarations.
Checked bit-for-bit compatibility with toolstash + buildall.
Change-Id: I601101d8177894adb9b0e3fb55dfe0ed4f544716
Reviewed-on: https://go-review.googlesource.com/6517
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Aconv is the pretty-printer for instruction opcodes like AMOVQ.
There was one for each architecture.
Make the space of A names have a different region for each architecture,
much as we did for the registers, so a single global Aconv function can
do the work. Each architecture registers its region as a slice of names
at a given offset.
The global names like CALL and JMP are now defined only once.
The A values are used for indexing tables, so make it easy to do the
indexing by making the offset maskable.
Remove a bunch of now-duplicated architecture-specific code.
Change-Id: Ib15647b7145a1c089e21e36543691a19e146b60e
Reviewed-on: https://go-review.googlesource.com/6620
Reviewed-by: Russ Cox <rsc@golang.org>
Run-TryBot: Rob Pike <r@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
It is unused and should have been deleted when Rconv was made
a global function.
Change-Id: Id745dcee6f0769604cabde04887c6d0c94855405
Reviewed-on: https://go-review.googlesource.com/6521
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Have the implementations of each architecture declare the one-operand,
destination-writing instructions instead of splitting the information between
there and asm.
Change-Id: I44899435011a4a7a398ed03c0801e9f81cc8c905
Reviewed-on: https://go-review.googlesource.com/6490
Reviewed-by: Russ Cox <rsc@golang.org>
Run rsc.io/grind rev a26569f on C->Go conversions.
The new change in grind is the inlining of goto targets.
If code says 'goto x' and the block starting at label x is unreachable
except through that goto and the code can be moved to where
the goto is without changing the meaning of its variable names,
grind does that move. Simlarly, a goto to a plain return statement
turns into that return statement (even if there are other paths to
the return statement).
Combined, these remove many long-distance gotos, which in turn
makes it possible to reduce the scope of more variable declarations.
(Because gotos can't jump across declarations, the gotos were
keeping the declarations from moving.)
Checked bit-for-bit compatibility with toolstash + buildall.
Reduces compiler runtime in html/template by about 12%.
Change-Id: Id727c0bd7763a61aa22f3daa00aeb8fccbc057a3
Reviewed-on: https://go-review.googlesource.com/6472
Reviewed-by: Aram Hăvărneanu <aram@mgk.ro>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Replaced by Ctxt.ByteOrder, which uses the standard binary.ByteOrder type.
Change-Id: I06cec0674c153a9ad75ff937f7eb934891effd0b
Reviewed-on: https://go-review.googlesource.com/6450
Reviewed-by: Dave Cheney <dave@cheney.net>
Reviewed-by: Rob Pike <r@golang.org>
Each architecture had its own Dconv (operand printer) but the syntax is
close to uniform and the code overlap was considerable. Consolidate these
into a single top-level function. A similar but smaller unification is done
for Mconv ("Name" formatter) as well.
The signature is changed. The flag was unused so drop it. Add a
function argument, Rconv, that must be supplied by the caller.
TODO: A future change will unify Rconv as well and this argument
will go away.
Some formats changed, because of the automatic consistency
created by unification. For instance, 0(R1) always prints as (R1)
now, and foo+0(SB) is just foo(SB). Before, some made these
simplifications and some didn't; now they all do.
Update the asm tests that depend on the format.
Change-Id: I6e3310bc19814c0c784ff0b960a154521acd9532
Reviewed-on: https://go-review.googlesource.com/5920
Reviewed-by: Russ Cox <rsc@golang.org>
Reconvert using rsc.io/c2go rev 27b3f59.
Changes to converter:
- fatal does not return, so no fallthrough after fatal in switch
- many more function results and variables identified as bool
- simplification of negated boolean expressions
Change-Id: I3bc67da5e46cb7ee613e230cf7e9533036cc870b
Reviewed-on: https://go-review.googlesource.com/5171
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
First draft of converted Go compiler, using rsc.io/c2go rev 83d795a.
Change-Id: I29f4c7010de07d2ff1947bbca9865879d83c32c3
Reviewed-on: https://go-review.googlesource.com/4851
Reviewed-by: Rob Pike <r@golang.org>
- obj: add a missing setting of the context for a generated JMP instruction
- asm: correct the encoding of mode (R)(R*scale)
- asm: fix a silly bug in the test for macro recursion.
- asm: accept address mode sym(R)(R*8); was an oversight
Change-Id: I27112eaaa1faa0d2ba97e414f0571b70733ea087
Reviewed-on: https://go-review.googlesource.com/4502
Reviewed-by: Russ Cox <rsc@golang.org>
cmd/internal/obj reconverted using rsc.io/c2go rev 2a95256.
- Brings in new, more regular Prog, Addr definitions
- Add Prog* argument to oclass in liblink/asm[68].c, for c2go conversion.
- Update objwriter for change in TEXT size encoding.
- Merge 5a, 6a, 8a, 9a changes into new5a, new6a, new8a, new9a (by hand).
- Add +build ignore to cmd/asm/internal/{addr,arch,asm}, cmd/asm.
They need to be updated for the changes.
- Reenable verifyAsm in cmd/go.
- Reenable GOOBJ=2 mode by default in liblink.
All architectures build successfully again.
Change-Id: I2c845c5d365aa484b570476898171bee657b626d
Reviewed-on: https://go-review.googlesource.com/3963
Reviewed-by: Rob Pike <r@golang.org>
cmd/internal/obj reconverted using rsc.io/c2go rev 40275b8.
All Prog*s need Ctxt field set so that the printer can tell
which architecture the Prog belongs to.
Use ctxt.NewProg consistently for this.
Change-Id: Ic981b3d68f24931ffae74a772e83a3dc2fdf518a
Reviewed-on: https://go-review.googlesource.com/3152
Reviewed-by: Rob Pike <r@golang.org>
