Mercurial > hg > index.cgi
view docs/internals.txt @ 418:3832a68d83ef
Fix internal compiler error on "var2 = var1 + 1" patterns
This appears to be the correct fix. It was provided by Tormod Volden
(debian.tormod@gmail.com). The final commit is substantially different from
Tormod's submission mostly due to housecleaning (removing the old patches
and updating the README). Tormod's comments follow.
The original addhi_mem_1 "insn" instruction pattern /matches/ two
memory operands, just with the /constraint/ that these are the same
location. A pattern match tells the compiler "you should be able to use
this, but you might have to work on it to meet the constraints". For
typical constraints on registers the compiler can add "reloads", moving
stuff between registers or from memory, until the constraints are met
and the instruction can be used. However, in this case, no amount of
reloads can make two memory locations the same if they already weren't,
so the compiler breaks down and cries "unable to generate reloads".
It seems this issue only appears if optimization is enabled. The proof
is in gcc's reload.c and is left as an exercise to the reader.
Limiting the matching pattern to identical memory operands avoids
these situations, while allowing the common "var++" cases.
References:
The pattern/constraints difference is explained in
https://gcc.gnu.org/onlinedocs/gccint/Simple-Constraints.html#index-other-register-constraints-3335
| author | William Astle <lost@l-w.ca> |
|---|---|
| date | Tue, 29 Mar 2016 21:21:49 -0600 |
| parents | 2c24602be78f |
| children |
line wrap: on
line source
LWASM Internals =============== LWASM is a table-driven assembler that notionally uses two passes. However, it implements its assembly in several passes as follows. Pass 1 ------ This pass reads the entire source code and parses each line into an internal representation. Macros, file inclusions, and conditional assembly instructions are resolved at this point as well. Instructions with known sizes will have their sizes resolved at this point. Pass 2 ------ Check all exported symbols for validity and set them as imports if the assembler state says so. Also resolve all symbol references in all expressions to be direct references either to the symbol table or to the import list. Pass 3 ------ This pass resolves all instruction sizes that can be resolved without forcing any instruction sizes. This pass will run repeatedly until no no new resolution occurs. Pass 4 ------ Work through all un-resolved instructions and force sizes. After each size is forced, try re-resolving all other instructions. This is done starting at the beginning of the source and working forward. If any instruction does not resolve when forced, an error will be thrown. Pass 5 ------ Constantize all line addresses and throw errors if any cannot be. This pass will repeat until no further lines addresses are reduced to constants at which time all lines will be checked for constant-ness. Pass 6 ------ Finalize all expressions related to instructions. Carp about any that cannot be reduced to a usable form. That means, for the non-object target all expressions must resolve to a constant. For the object form, all expressions must resolve to symbol references and constants. Those symbol references may be internal or external. Pass 7 ------ Emit object code for each line for later output.