summary refs log tree commit diff
AgeCommit message (Collapse)Author
2019-02-21doc: Include `align` in data BNFMichael Forney
2019-02-21Fix typoMichael Forney
2019-02-18mark phi arguments as escapingQuentin Carbonneaux
Thanks to Michael Forney for spotting this oversight and providing the test case. Note: because esc() leaves ABot unchanged, the assertion "a->type == ABot" on line 122 remains valid.
2019-02-06soften heuristic of 316b57Quentin Carbonneaux
Instead of systematically spilling any temp live in an exit branch but not in the part of the loop already processed, only spill when it is already known to have been spilled.
2019-02-062 bug fixes in regaQuentin Carbonneaux
The worst one was that "part 3" of rega() could break the critical invariant that two interferring temporaries get assigned different registers. This is fixed by being careful when changing the register of a temporary based on predecessor blocks. Thanks to Michael Forney for reporting these bugs and helping with the analysis.
2019-02-05fix a bad bug in regalloc boilerplateQuentin Carbonneaux
That was silly... I believe qbe still managed to work because bitsets are only used inside a basic block where rcopy() is not used.
2019-02-05new spiller heuristic for loopsQuentin Carbonneaux
If a variable is spilled in a loop, the spiller now tries to keep it spilled over the whole loop. Thanks to Michael Forney for sharing a test case exhibiting a pathological reload.
2018-04-26more compiler warnings...Quentin Carbonneaux
2018-04-26Fix compiler warnings.Emil Skoeldberg
Compiler warned about comparison between signed and unsigned values.
2017-10-07fix compiler command in testccEugene Sharygin
This commit adds missing quotation marks around the argument to the function, and changes the value of `-x' option to `c` (lowercase) as per GCC manual [1]. [1]: https://gcc.gnu.org/onlinedocs/gcc-7.2.0/gcc/Overall-Options.html
2017-09-25adjust test.sh for ubuntuQuentin Carbonneaux
2017-09-22mark printf call as variadic in testQuentin Carbonneaux
2017-08-17fix bug in jumps simplificationQuentin Carbonneaux
In presence of jump loops, the algorithm would create cycles in the disjoint-set data structure. This caused infinite recursion and stack overflows.
2017-07-30fix dynamic stack allocs for amd64Quentin Carbonneaux
The arm64 might have the same problem but it is currently unable to handle them even in instruction selection. Thanks to Jean Dao for reporting the bug.
2017-06-06fix fp subtractions on amd64Quentin Carbonneaux
The stashing of constants in gas.c was also changed to support 16-bytes constants.
2017-06-06isreg() does not need to be inlinedQuentin Carbonneaux
2017-06-06fix floating-point divisionQuentin Carbonneaux
It never worked until today.
2017-05-17free the typ vector at the end of parse()Quentin Carbonneaux
2017-05-17intern symbol namesQuentin Carbonneaux
Symbols in the source file are still limited in length because the rest of the code assumes that strings always fit in NString bytes. Regardless, there is already a benefit because comparing/copying symbol names does not require using strcmp()/strcpy() anymore.
2017-05-16new hinting in the register allocatorQuentin Carbonneaux
The previous heuristics were ad hoc and it was hard to understand why they worked at all. This patch can be summarized in three points: 1. When a register is freed (an instruction assigns it), we try to find if a temporary would like to be in it, and if we find one, we move it in the newly freed register. I call this an "eager move". 2. Temporaries now remember in what register they were last allocated; this information is stored in the field Tmp.visit, and prevails on the field Tmp.hint when it is set. (This makes having the same hint for interfering temporaries not so disastrous.) 3. Blocks are now allocated in "onion" order, from the innermost loop to the outermost. This is the change I am the least sure about; it should be evaluated thorougly.
2017-05-16change the computation of spill costs for phisQuentin Carbonneaux
I now take the view that a phi is "used" at the end of all the predecessors. (Think that copies are made to phis at the end of all predecessors.)
2017-05-16do not account for interferences in phi classesQuentin Carbonneaux
Before this commit, I tried to make sure that two interfering temporaries never ended up in the same phi class. This was to make sure that their register hints were not counterproductively stepping on each other's toes. The idea is fine, but: * the implementation is clumsy because it mixes the orthogonal concepts of (i) interference and (ii) phi classes; * the hinting process in the register allocator is hard to understand because the disjoint-set data structure used for phi classes is cut in arbitrary places. After this commit, the phi classes *really* are phi classes represented with a proper disjoint-set data structure.
2017-04-26Small corrections in documentationQuentin Rameau
2017-04-18documentation updateQuentin Carbonneaux
2017-04-16minor changes for env parameterQuentin Carbonneaux
Since the environment can only be of type `l`, do not require to write it down. I also tightened the type information of the `pare` and `arge` instructions.
2017-04-14remove html converterQuentin Carbonneaux
2017-04-11unscrew freebsd testsQuentin Carbonneaux
2017-04-11simplify amd64 aggregates classificationQuentin Carbonneaux
2017-04-10bump the size of the instruction bufferQuentin Carbonneaux
Ori needs this. It should not cost much more memory at runtime, only a minimal amount of address space.
2017-04-10simplify slot logic in alias analysisQuentin Carbonneaux
The previous code was buggy. It would put a stack pointer on the heap when handling "add $foo, 42". The new code is more straightforward and hopefully more correct. Only temporaries with a "stack" alias class will have a slot pointer.
2017-04-09always disable pie in testsQuentin Carbonneaux
2017-04-08misc fixes for osxQuentin Carbonneaux
With the default toolchain, it looks like we have to make sure all symbols are loaded using rip-relative addressing.
2017-04-08add handy src target to the MakefileQuentin Carbonneaux
Intended usage: c_count `make src` (or similar).
2017-04-08add instructions to build on windowsQuentin Carbonneaux
2017-04-08nits in the documentationQuentin Carbonneaux
2017-04-08use amd64 instead of x64 in abi docQuentin Carbonneaux
2017-04-08new abi test for arm64 HFAsQuentin Carbonneaux
2017-04-08enable arm64 testsQuentin Carbonneaux
The vararg tests had to be changed because va_list is 32-bit wide on arm. The astute reader will notice that the way we pass va_list values is wrong, we should be using the ':valist' type as defined below instead of 'l'. But eh, that works for now, because of the ABI. type :valist = align 8 { 32 }
2017-04-08add cross testing for arm64Quentin Carbonneaux
2017-04-08new arm64 backend, yeepeeQuentin Carbonneaux
2017-04-08rework storage of typesQuentin Carbonneaux
The arm64 ABI needs to know precisely what floating point types are being used, so we need to store that information. I also made typ[] a dynamic array.
2017-04-08prepare for multi-targetQuentin Carbonneaux
This big diff does multiple changes to allow the addition of new targets to qbe. The changes are listed below in decreasing order of impact. 1. Add a new Target structure. To add support for a given target, one has to implement all the members of the Target structure. All the source files where changed to use this interface where needed. 2. Single out amd64-specific code. In this commit, the amd64 target T_amd64_sysv is the only target available, it is implemented in the amd64/ directory. All the non-static items in this directory are prefixed with either amd64_ or amd64_sysv (for items that are specific to the System V ABI). 3. Centralize Ops information. There is now a file 'ops.h' that must be used to store all the available operations together with their metadata. The various targets will only select what they need; but it is beneficial that there is only *one* place to change to add a new instruction. One good side effect of this change is that any operation 'xyz' in the IL now as a corresponding 'Oxyz' in the code. 4. Misc fixes. One notable change is that instruction selection now generates generic comparison operations and the lowering to the target's comparisons is done in the emitter. GAS directives for data are the same for many targets, so data emission was extracted in a file 'gas.c'. 5. Modularize the Makefile. The Makefile now has a list of C files that are target-independent (SRC), and one list of C files per target. Each target can also use its own 'all.h' header (for example to define registers).
2017-04-08fix bug in union size computationQuentin Carbonneaux
The size of a union is the size of the largest element aligned with the largest alignment. For example, the size of the following union is 16, not 13 (as returned before this patch). union { char c[13]; int i; };
2017-03-29improve global registers handlingQuentin Carbonneaux
The register allocation now has stricter assertions about global registers. The stricter assertions required changes in the spiller: We now correctly indicate to the register allocator what registers are used by "ret" instructions.
2017-03-03add another idiomatic C test (rega does no good)Quentin Carbonneaux
2017-02-27make install/uninstall phoniesQuentin Carbonneaux
2017-02-27add install and uninstall targets to MakefileQuentin Rameau
2017-02-27remove unused parameter from uffind()Quentin Carbonneaux
2017-02-27minor tweaks to simpljmp passQuentin Carbonneaux
2017-02-27fix int parsingQuentin Carbonneaux
The spec says that numbers can be arbitrarily big, and only the last 64 bits will be taken into consideration. Calling sscanf does not implement this, so I wrote an ad-hoc function.