| 2005-04-08 |
Linus Torvalds | Make range check code a bit more readable (and more...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add compile-time "range-check" infrastructure to sparse
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make sure we evaluate pointer comparisons as unsigned.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Shut up informational messages once we've hit the maximum
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make "check" (aka "sparse") check data-dropping casts.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make simplification remove casts that change neither...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Split OP_CAST into signed, unsigned and FP casts.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Simplify OP_CAST of OP_AND.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Don't optimize away casts too early.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Duh. When a function returns VOID, we should _not_...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Fix code generation confusion between OP_ADDR and the
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make the example code generator do something half-way...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make constant instruction simplification take the sign...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Split the binops where signedness matters into unsigned...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make linearizer able to handle assignment ops where the
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Evaluate asm outputs as assignments.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | We didn't mark symbols with their address taken as...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make expression type enums start at one..
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | inlining: don't copy an initializer expression for...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Linearize EXPR_POS initializers more carefully: they can
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add option "-Wptr-subtraction-blows" to warn about...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make "check_access()" take the size of the access properly
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Don't do the pointer offset update for a pointer addition
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Don't try to recursively check for an associative insn.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add warning for accessing outside of a symbol
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Simplify OP_PTRCAST ops too for now.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Fix array size calculation when the last entry is an...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make output_insn() tell where it was called from, and...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make "fill_reg" do somewhat the right thing when we...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Oops. Missed a place where we still tested for "busy...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make the argument storage setup be a bit more accurate.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Toplevel symbols are externally addressable only if...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make "reg->busy" mean how many "operands" actually...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Start moving to a more symbol "struct operand" notion...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Fix "address_taken()" function to work at least pitifully
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Get comparison sizes right.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Split OP_SETVAL into OP_SETVAL (fp expressions and...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make sure to mark all registers that have already been...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | The stack offset is global, not per-bb.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Duh. We marked the wrong register REG_FIXED when we...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add some back-of-the-envelope support for asm inputs...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Do absolutely horrid job of generating code for asms.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Use the one-deep CC-cache for OP_SEL too.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Simplify conditional on cast-to-larger-size to conditional...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | If we decide to mark a register as being its own storage,
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add a one-deep CC-cache for condition code setting...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Verify that output/input asm constraints really look...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Save off the asm parameter name too.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Show asm inputs/outputs as bugus instructions as opposed...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Start looking at asms in code generation.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make asm linearization not drop the constraints.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Do some kind of signed cast too.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Do slightly better on casts.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Teach code generator about commutative operations.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Oops. Didn't remember about BINCMP ops when doing the...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Teach simplification about associative operators.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make commutative operations use a canonical order.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Mark the backing store storage dead when marking a...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make "find_pseudo_storage()" return the storage hash...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Split up the code that finds the underlying storage...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add storage usage verification phase.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add support for various arch-specific storage allocation
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add the argument pseudos to the "enter" instruction
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Expose lookup_storage/add_storage to code generator.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Sort switch target list
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Kill off dead pseudos before doing target allocation...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Keep dead pseudos in the register "busy" count, add...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add comment on where incoming pseudos come from.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Generate pseudo-code for OP_SEL.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make target register allocation prefer empty registers.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Generate cheesy "cast" instructions.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add fake OP_CALL code generation.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Be a bit more forgiving about impossible output register...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Output unimplemented instructions as such, rather
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Clean up final register flushing, and allow flushing...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add comment about input storage selection.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | When fulfilling the output register requirements, don't
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Don't kill the same pseudo twice.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Remove pseudos from liveness list when they are defined.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Track phi uses in a separate pass from the liveness...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Clean up linearizer output that got uglified by the...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add tons more comments about register state changes...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make "remove_pseudo()" return whether it removed a...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | When removing/replacing pointer list entries, return...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | PHI pseudos aren't supposed to show up on the liveness
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Remove old pseudos from register list when we redefine one
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make pretty helper functions for showing individual...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Don't re-use registers for several different output...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Use the ptr-list tagging for marking pseudos dirty.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Support tagged add_ptr_list
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Be smarter about when we need to flush a pseudo.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Now that we can have multiple pseudos per reg, we should
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Generate some kind of code for OP_RET
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Remove stale assertion.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Make the hardreg pseudo pointer be a _list_ of pseudos...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add macros to set/read the pointer list tags.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Beginning infrastructure for tagged lists.
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Move remove_pseudo() to linearize.h
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Teach register "allocator" about preferred register...
|
commit | commitdiff | tree |
| 2005-04-08 |
Linus Torvalds | Add "stream_name()" helper function, and use it.
|
commit | commitdiff | tree |
