If X < 0, then lowering the addi might cause the code to use the stack
space before allocating it. This is a bug because an asynchronous
signal handler can overwrite the unallocated stack space.
Tests pass if one edits the top build.lua to uncomment "qemuppc" from
both vars.plats and vars.plats_with_tests, and one leaves mcg in
plat/qemuppc/descr.
Add or correct some EM instructions in treebuilder.c:
- "lof", "stf": handle negative offsets in load() and store().
- "cuu": add using IR_FROMUI.
- "lim", "sim": keep an entire word in ".ignmask", to be compatible
with mach/powerpc/libem/trp.s and ncg. We also keep a word in
".ignmask" in ncg for both i386 and m68020.
- "trp": pass trap number in register. See comment in
helper_function_with_arg().
- "sig": push the old value of .trppc on the stack.
- "and ?", "ior ?", "xor ?", "com ?", "cms ?", "set ?", "inn ?":
connect to helper functions in libem.
- "blm", "bls": drop call to memmove() and use new helper ".bls4",
because tests/plat/structcopy_e.c can't call memmove().
- "xor s", "cms s": if s is large, fall back on helper function.
- "rol", "ror": add by decomposing each rotate into 4 IR ops.
- "rck s", "bls s": make fatal unless s is word size.
- "loi": push multiple loads in the correct order.
- "dup s", "exg s": if s is large, fall back on helper.
- "dus": add using new helper ".dus4".
- "lxl", "lxa": follow the static chain, not the dynamic chain.
- "lor 1": materialise the stack before pushing the stack pointer.
- "lor 2", "str 2": make fatal.
- "los", "sts": drop calls to memcpy() and use helpers ".los4" and
and ".sts4", so lang/m2/libm2/LtoUset.e starts working.
- "gto": correctly read descriptor.
Change mach/powerpc/mcg/table:
- ANY.L: add for "asp -8".
- LOAD.L: work around register corruption.
- COMPAREUL.I: add for "cms 8".
In the instruction list, put /* kills xer */ for sraw, srawi, subfic;
and correct the (now unused) "addi." and "lfdu".
Change MACHOPT_F from -m3 to -m2. This changes the code for 15 * i
from
slwi r3,r4,4
subfic r5,r4,0
add r3,r3,r5
to
mulli r3,r4,15
If the sequence "slwi subfic addi" takes 3 cycles and 12 bytes, and
mulli takes 3 cycles and 4 bytes, then mulli is better.
Copy and adapt code from mach/{i386,m68020}/ncg/mach.c to pass the
debugging stabs from EM to assembly. The next tools (as, led, cv)
already know how to put the stabs in the Mach-o executable.
Modify the function prolog/prologue so gdb uses fp, not sp, for N_LSYM
and N_PSYM stabs. Simplify prolog() by reducing differences between
stabs and no stabs, and zero and nonzero framesize. For files without
stabs, the new prolog has the same number of instructions and memory
accesses as the old prolog, and to run at about the same speed on my
PowerPC Mac.
This is enough to see some info for global and local variables in gdb
for Mac OS X. I still can't get a backtrace; gdb gets confused
because EM and ncg don't link 0(sp) to the previous stack frame.
I don't expect `ack -mlinuxppc -g` to work with gdb for Linux, because
we prepend underscores to the symbol table, which is correct for
Mach-o but wrong for ELF.
This simplifies parts of the PowerPC table and causes ncg to better
decide whether to push sp or fp to the real stack, or coerce it to
REG3, or coerce it to REG-REG3, or move it to a regvar. These better
decisions remove extra _mr_ instructions.
The idea comes from mach/powerpc/arm/table, where SP has a property
STACKPOINTER and LB has LOCALBASE. I don't need two properties, so I
make one property SPFP for both registers.
When I wrote fef 8, I forgot to test denormalized numbers. Oops. Now
fix two of my mistakes:
- When checking for zero, `extrwi r6, r3, 22, 12` needs to be
`extrwi r6, r3, 20, 12`. There are only 20 bits to extract.
- After the multiplication by 2**64, I forgot to put the fraction in
[0.5, 1) or (-1, 0.5] by setting IEEE exponent = 1022.
Teach fif 8 about signed zero and NaN.
In ncg/table, change cmf so NaN is not equal to any value, and comment
why ordered comparisons don't work with NaN. Also add cost for
fctwiz, remove extra `uses REG`.
Edit comment in cfu8.s because the conditional branch might be before
or after fctwiz.
With my PowerBook G4, a program that converts values from 1.0 to
4000000.0 runs in about 0.32s with the old .cfu8 and 0.29s with this
shrunken .cfu8
Leave a comment about other ways to implement .cfu8
This reduces code size, because ncg emits too many "addi sp, sp, X"
instructions when unstacking things. Now top lowers "addi sp, sp, X"
by lifting other instructions. This sometimes creates chances to
merge or delete _addi_ instructions. If no such chance is found, the
_addi_ remains uselessly lowered.
Edit ncg/table to remove something that top now does.
Edit ncg/mach.c to remove some spaces after commas. This removes a
whitespace difference between *.s and *.so files, because top removes
the space.
_lim_ must use _loe_ (load word external), not _lde_ (load double-word
external).
The new patterns for _lxl_, _lxa_, _lor_, _str_ emit shorter code in
some cases. The change from GPR_EXPR to REG_EXPR allows moving
LXFRAME to a register variable.
Add more "reusing" clauses. We have enough registers that ncg almost
never reuses a register, but sometimes it can reuse r3.
In mach.c, emit one fewer instruction in procedures with no locals.
Fix PowerPC ncg so setjmp() returns the correct value. I got unlucky
when ncg picked r3 for "uses REG"; this destroyed the return value in
r3 and caused the new test to fail.
The new test rck_e.e segfaults on PowerPC unless I make some changes.
The inline code for _rck_ was wrong because it didn't allow the trap
handler to return. _sig_ forgot to push the old trap handler.
Move plat/linuxppc/libsys/trap.s to mach/powerpc/libem/trp.s and
rewrite it with simplified/extended mnemonics. Remove .trap alias for
.trp procedure. Add a missing `mtspr lr, r0` so we can return from
the trap handler. Call write() and _exit() so trp.s works with both
linuxppc and osxppc. Before, Mac OS X was wrongly using the trap.s
for Linux.
In powerpc/libem, simplify .aar4; teach .csa and .csb to raise the
trap if the default target is zero.
C programs don't need these changes. You may relink your C programs
with the changed .csa and .csb, but C code doesn't raise the trap.
Modula-2 code can raise traps, so you may want to relink your Modula-2
programs with the changed libem, but you might keep your old .o files
from Modula-2. You may need to recompile your Pascal programs (delete
old .o files from Pascal) because the Pascal compiler might use _rck_.
I understand `loi 4` more easily than `loi INT32`, because `loi 4`
appears in .e files. So remove INT8, INT16, INT32, INT64.
Add a comment to explain r3 during unconditional jumps.
When storing to a local, stop killing the tokens of other locals,
unless they might overlap with the stored local. This helps some
procedures that juggle locals when the locals aren't in registers.
Also use FRAME_V tokens for locals in statically enclosing procedures.
Rewrite _lxa_ as _lxl_, to skip the `addi ?,?,8` if we can add 8 to
the next constant. The PowerPC code from _lxl_ is now sometimes
better, sometimes worse than before.
The i386 table provided the idea to use %size to find overlapping
locals.
ncgg has parsed the optional conditional expression (optexpr) of each
splitting coercion since commit 72b83cc in 1985; but for almost 33
years, ncg has ignored the expression in c2_expr.
Few tables had conditional coercions (I only found them in arm and
m68020), and no tables had conditional splitting coercions, so this
only becomes a problem now as I try to add a conditional splitting
coercion to powerpc.
Prefer the rule with REG FREG, by coercing IND_RC_D or IND_RR_D to
FREG. This rule looks better to ncg. When ncg chose between coercion
to REG IND_RC_D or coercion to REG FREG, it chose REG FREG. It only
chose REG IND_RC_D if the stack had exact REG IND_RC_D.
Allow asp 4, exg 4 to shuffle tokens without coercing them into
registers; but comment why dup 4, dup 8 coerce tokens into registers.
Allow dup, dus, exg with larger sizes; and add tests dup_e.e and
exg_e.e to check that dup 20, dus, exg 20 work as well in powerpc as
in i80 and i86.
Then powerpc failed to compile loc 2 loc 4 cuu in dup_e.e. Revise the
integer conversions, so powerpc can compile and pass the test.
When a rule `uses REG ... yields %a`, the result %a is always a
temporary, never a regvar. If the EM code uses _stl_ to put the
result in a regvar, then ncg emits _mr_ to move %a to the regvar.
There are two ways to put the result in the regvar without %a:
1. Yield a token, as in `yields {MUL_RR, %2, %1}`, so that _stl_
can move the token to the regvar without using %a.
2. Provide a pattern, like `sli stl`, that just puts the result
in `{LOCAL, $2}` and not %a.
Allow some tokens, like SUM_RIS and XEQ, onto the stack; and add
tokens like MUL_RR, and patterns like `sli stl`.
Delete patterns for `stl lol` and `sdl ldl` to avoid an extra
temporary %a when the local is a regvar. Delete `lal sti lal loi`
because it would emit wrong code.
Rename token CONST to C. Define set CONST = C + CONST_STACK. The
instructions with CONST operands can now accept CONST_STACK tokens;
some cases of {CONST, %1.val} become %1.
Also simplify two of _rlwinm_ into _slwi_ and _srwi_.
EM instructions _rol_ and _ror_ do rotate an integer left or right.
Our compilers and optimizers never emit _rol_ nor _ror_, but I might
want to use them in the future.
Add _rol_ and _ror_ to powerpc. Fix `rol 4` and `ror 4` in both i80
and i86, where the rules for `rol 4` and `ror 4` seem to have never
been tested until now.
The code used `sphl` to set the stack pointer, but the correct value
was in de, not hl. Fix by swapping the values of de and hl, so `sphl`
is now correct. When we shrink an integer from 4 to 2 bytes, both
registers de and hl point to copies of the result, but only one
register preserves the stack below the result.
This fixes writehex() in tests/plat/lib/test.c, when I compile it with
ack -mcpm, so it preserves the pointer to "0123456789abcdef", so it
writes hexadecimal digits and not garbage.
This bug goes back to commit 157b243 of Mar 18, 1985, so the bug is
32 years old, and probably the oldest bug that I ever fixed.
I compiled tests/plat/lib/test.c with ack -mcpm, but i80 ncg did emit
wrong code in writehex(uint32_t) for
"0123456789abcdef"[code & 0xf]
The code called '.and' to evaluate `code & 0xf`, then tried to call
'.cii' to narrow the result from 4 to 2 bytes, but it passed garbage
instead of 4 to '.cii'. The rule for '.and' was
pat and defined($1)
kills ALL
uses dereg={const2,$1}
gen Call {label,".and"}
This failed to kill register de={const2,4}, so ncg pushed de,
expecting to push 4, but actually pushing garbage.
Fix such rules using `mvi a,...` or `lxi de,...` so ncg doesn't track
the token in the register. This is like the i86 table. A different
fix would use a dummy instruction `killreg a` or `killreg de` like the
m68020 table.
Also correct 1 to $1 when calling '.exg'.
gcc gave an error because the `char *` parameter doesn't match the
`const char *` in the prototype of regsave(). clang didn't give an
error. I added the prototype in commit 5301cce.
This breaks all machines because the declared return type void
disagrees with the implicit return type int (when I compile mach.c
with clang). Unbreak i386, i80, i86, m68020, powerpc, vc4 by adding
the return types to mach.c. We don't build any other machines; they
are broken since commit a46ee91 (May 19, 2013) declared void prolog()
and commit fd91851 (Nov 10, 2016) declared void mes(), with both
declarations in mach/proto/ncg/fillem.c.
Also fix mach/vc4/ncg/mach.c where type full is long, so fprintf()
must use "%ld" not "%d" to print full nlocals.
Files that #include "equiv.h" must do so after including "data.h", now
that a function prototype in equiv.h uses type rl_p from data.h.
Adjust style, changing some `for(...)` to `for (...)`. The style in
mach/proto/ncg is less than consistent; the big annoyance now is that
some files want tabs at 4 spaces, others want tabs at 8 spaces.
Put the declarations in "data.h", because that header declares the
types cost_t and token_p. Also #include <cgg_cg.h> from "data.h" to
get types c3_p and set_p, and guard <cgg_cg.h> against multiple
inclusion.
*Important:* You must "make clean" after checking out this commit,
because the build had copied the old "assert.h" to several places in
obj/. If you don't "make clean", then the compiler finds the old
"assert.h" before libc <assert.h>, and the build fails because this
commit removes badassertion() in subr.c. After "make clean", the
compiler finds libc <assert.h> and the build succeeds.
Switch from custom assert() to libc assert() in mach/proto/as.
Continue to disable asserts if DEBUG == 0.
This change found a problem in the build system; comm2.y was missing
depedencies on comm0.h and comm1.h. Add the missing dependencies to
the cppfile rule. Allow the dependencies by modifying cppfile in
first/build.lua to act like cfile if t.dir is false.
Now that comm2.y gets rebuilt, I must fix the wrong prototype of
yyparse() in comm1.h.
I got unlucky as induo() in comm5.c was reading beyond the end of the
array. It found an operator "= " ('=' then space) in the garbage, so
it returned a garbage token number, and "VAR = 123" became a syntax
error. Unbreak induo() by terminating the array.
Change "register i;" to "int i;" to so clang stops warning about
implicit int. Use function prototypes so clang stops warning about
implicitly declared functions.
In util/ncgg, add two more errors for tables using reglap:
- "Two sizes of reg_float can't be same size"
- "Missing reg_float of size %d to contain %s"
In mach/proto/ncg, rename macro isregvar_size() to PICK_REGVAR(), so
the macro doesn't look like a function. This macro sometimes doesn't
evaluate its second argument.
In mach/powerpc/ncg/mach.c, change type of lfs_set to uint32_t, and
change the left shifts from 1U<<regno to (uint32_t)1<<regno, because
1U would be too small for machines with 16-bit int.
This relocation is specific to PowerPC. @davidgiven suggested the
name RELOPPC_LIS in
https://github.com/davidgiven/ack/pull/52#issuecomment-279856501
Reindent the list in h/out.h and util/led/ack.out.5 because
RELOLIS_PPC is a long name. I use spaces and no tabs because the tabs
looked bad in the manual page.
Because lwzu or stwu moves the pointer, I can remove an addi
instruction from the loop, so the loop is slightly faster.
I wrote a benchmark in Modula-2 that exercises some of these loops. I
measured its time on my old PowerPC Mac. Its user time decreases from
8.401s to 8.217s with the tighter loops.
The result of single-precision fadds, fsubs, and such can go into a
register variable, like we already do with double precision. This
avoids an extra fmr from a temporary register to the regvar.
Do the conversion by calling .cif8 or .cuf8 in libem, as it was done
before my commit 1de1e8f. I used the inline conversion to experiment
with the register allocator, which was too slow until c5bb3be.
Now that libem has the only copy of the code, move some comments and
code changes there.
Rename GPRE to GPR_EXPR, then define FPR_EXPR and FSREG_EXPR. Use
them for moves to register variables.
Keep "kills regvar($1)", because deleting it and recompiling libc
would cause many failures in my test programs. Add comment to warn,
/* ncg fails to infer that regvar($1) is dead! */
Remove "kills LOCAL %off==$1" because it seems to have no effect.
If the ncg table uses reglap, then regvar($1, reg_float) would have
two sizes of registers. An error from ncgg would happen if regvar()
was in a token that allows only one size. Now one can pick a size
with regvar_w() for word size or regvar_d() for double-word size.
Add regvar_d and regvar_w as keywords in ncgg. Modify EX_REGVAR to
include the register size. In ncg, add some checks for the register
size. In tables without reglap, regvar() works as before, and ncg
ignores the register size in EX_REGVAR.
After the RA phase of ego, a procedure may put single-word and
double-word values in the same reg_float. Then ncg will use both
LOCAL and DLOCAL tokens at the same offset.
I add isregvar_size() to ncg. It receives the size of the LOCAL or
DLOCAL token, and picks the register of the correct size. This fixes
a problem where ncg got the wrong-size register and corrupted the
stack. This problem caused one of my test programs to segfault from
stack underflow.
Also adjust how fixregvars() handles both sizes.
This merges several fixes and improvements from upstream. This
includes commit 5f6a773 to turn off qemuppc. I see several failing
tests from qemuppc; this merge will hide the test failures.
The new feature "reglap" allows two sizes of floating-point register
variables (reg_float), if each register overlaps a single register of
the other size. PowerPC ncg uses reglap to define 4-byte instances
of f14 to f31 that overlap the 8-byte instances.
When ncgg sees the definition of fs14("f14")=f14, it removes the
8-byte f14 from its rvnumbers array, and adds the 4-byte fs14 in its
place. Later, when ncg puts a variable in fs14, if it is an 8-byte
variable, then ncg switches to the 8-byte f14. The code has
/* reglap */ comments in util/ncgg or #ifdef REGLAP in mach/proto/ncg
reglap became necessary because my commit a20b87c caused PowerPC ego
to allocate reg_float in both 4-byte and 8-byte sizes.
After c5bb3be, ncg began to allocate regvars from r13 up. I reorder
the regvars so ncg again allocates them from r31 down. I also reorder
the other registers.
This exposed a bug in my rule for ret 8. It was wrong if item %2 was
in r3, because I moved %1 to r3 before %2 to r4. Fix it by adding
back an individual register class for r3 (called REG3 here, GPR3 in
c5bb3be).
Also fix my typo in mach.c that made a syntax error in assembly.
This removes a wrong-way dependency of libsys on libem. The C
functions in libsys called .ret, but libsys is after libem in the
linker arguments, so the linker didn't find .ret unless something else
had called .ret. Almost everything called .ret, but I got a linker
error when I wrote an assembly program using the EM runtime, because
my assembly program didn't call .ret.
Add a dummy comment to build.lua, so git checkout touches that file,
the build system reconfigures itself, and the *.s glob sees that ret.s
has gone.
This is like David Given's change to util/ncgg in d89f172. I need
this change in mach/proto/ncg to see fatal messages, because a 64-bit
pointer doesn't fit in an int.
Use f14 to f31 as register variables for 8-byte double-precison.
There are no regvars for 4-byte double precision, because all
regvar(reg_float) must have the same size. I expect more programs to
prefer 8-byte double precision.
Teach mach/powerpc/ncg/mach.c to emit stfd and lfd instructions to
save and restore 8-byte regvars. Delay emitting the function prolog
until f_regsave(), so we can use one addi to make stack space for both
local vars and saved registers. Be more careful with types in mach.c;
don't assume that int and long and full are the same.
In ncg table, add f14 to f31 as register variables, and some rules to
use them. Add rules to put the result of fadd, fsub, fmul, fdiv, fneg
in a regvar. Without such rules, the result would go in a scratch
FREG, and we would need fmr to move it to the regvar. Also add a rule
for pat sdl inreg($1)==reg_float with STACK, so we can unstack the
value directly into the regvar, again without a scratch FREG and fmr.
Edit util/ego/descr/powerpc.descr to tell ego about the new float
regvars. This might not be working right; ego usually decides against
using any float regvars, so ack -O1 (not running ego) uses the
regvars, but ack -O4 (running ego) doesn't use the regvars.
Beware that ack -mosxppc runs ego using powerpc.descr but -mlinuxppc
and -mqemuppc run ego without a config file (since 8ef7c31). I am
testing powerpc.descr with a local edit to plat/linuxppc/descr to run
ego with powerpc.descr there, but I did not commit my local edit.
This fixes lxl 1 (so it follows the static chain, not the dynamic
chain) and provides lxl 2 and greater. The Modula-2 compiler uses lxl
for nested procedures, so they can access the variables of the
enclosing procedures.
I added REG_PAIR in cfbc537 to speed up the register allocator,
because ncg was taking about 2 seconds on each sti 8. I defined only
4 such pairs, so allocating REG_PAIR was much faster than allocating
REG REG.
After my last commit c5bb3be, allocation of REG REG is fast, and
REG_PAIR seems unnecessary.
The table for PowerPC had placed each GPR and FPR into an individual
register class (like GPR3, GPR4, FPR1, FPR2), and had used these
classes to coerce stack values into specific registers. But ncg does
not like having many register classes.
In http://tack.sourceforge.net/olddocs/ncg.pdf
Hans van Staveren wrote:
> Every extra property means the register set is more unorthogonal and
> *cg* execution time is influenced by that, because it has to take
> into account a larger set of registers that are not equivalent. So
> try to keep the number of different register classes to a minimum.
Recent changes to the PowerPC table have removed many coercions to
specific registers. Many functions in libem switched from taking
values in registers to taking them from the stack (see dc05cb2).
I now remove all 64 individual register classes of GPR and FPR. In
the few cases where I need a stack value in a specific register, I now
do a move (as the arm and m68020 tables do).
This commit speeds the compilation of some files. For my test file
fconv.c, the compilation time goes from over 20 seconds to under 1
second. My fconv.c has 4 conversions from floats to integers, and the
table has my experimental rules that do the conversions by allocating
4 or 5 registers.
Switch .cms to pass inputs and outputs on the real stack, not in
registers; like we do with .and, .or (81c677d) and .xor (c578c49).
At this point, nearly all functions in libem use the real stack, not
registers, for passing inputs and outputs. This simplifies the ncg
table (which needs fewer lists of specific registers) but slows calls
to libem.
For example, after ba9b021, each call to .aar4 is about 10
instructions slower. I moved 3 inputs and 1 output from registers to
the real stack. A program would take 4 instructions to move registers
to stack, 4 to move stack to registers, and perhaps 2 to adjust the
stack pointer.
Our libem had two implementations of loading a block from a stack, one
for lar 4 and one for los 4. Now lar 4 and los 4 share the code in
.los4. Likewise, sar 4 and sts 4 share the code in .sts4.
Rename .los to .los4 and .sts to .sts4, because they implement los 4
and sts 4. Remove the special case for loading or storing 4 bytes,
because we can do it with 1 iteration of the loop. Remove the lines
to "align size" where the size must already be a multiple of 4.
Fix the upper bound check in .aar4.
Change .aar4, .lar4, .los4, .sar4, .sts4 to pass all operands on the
real stack, except that .los4 and .sts4 take the size in register r3.
Have .aar4 set r3 to the size of the array element. So lar 4 is just
.aar4 then .los4, and sar 4 is just .aar4 then .sts4.
ncg no longer calls .lar4 and .sar4 in libem, because it inlines the
code; but I keep .lar4 and .sar4 in libem, because mcg references
them. They might or might not work in mcg.
Reorder the code in .fef8 and .fif8 so that in the usual case, we fall
through to the blr without taking any branches. The usual case, by my
guess, is .fef8 with normalized numbers or .fif8 with small integers.
I change .fef8 and .fif8 to pass values on the real stack, not in
specific registers. This simplifies the ncg table, and might help me
experiment with changes to the ncg table.
This change might or might not help mcg. Seems that mcg always uses
the stack to pass values to libem, but I have not tested .fef8 or
.fif8 with mcg.
Switch some conversions from libem calls to inline code. The
conversions from integers to floats are now too slow, because each
conversion allocates 4 or 5 registers, and the register allocator is
too slow. I might use these slow conversions to experiment with the
register allocator.
I add the missing conversions between 4-byte single floats and
integers, simply by going through 8-byte double floats. (These
replace the calls to nonexistant functions in libem.)
I remove the placeholder for fef 4, because it doesn't exist in libem,
and our language runtimes only use fef 8.
This replaces a call to memmove() in libc. That was working for me,
but it can fail because EM programs don't always link to libc.
blm and bls only need to copy aligned words. They don't need to copy
bytes, and they don't need to copy between overlapping buffers, as
memmove() does. So the new loop is simpler than memmove().
Remove one addi instruction from some loops. These loops had
increased 2 pointers, they now increase 1 index. I must initialize
the index, so I add "li r6, 0" before each loop.
Change .zer to use subf instead of neg, add.
Change .xor to take the size on the real stack, as .and and .or have
done since 81c677d.
Use extended "mr" instead of basic "or" to move registers. Both "mr"
and "or" encode the same machine instruction. With "mr", I can more
easily search the assembly output for register moves.
Fold several stacking rules into a single rule ANY_BHW-REG to STACK.
Remove the EM patterns for loc mlu $2==2 and loc slu. The first
pattern had the wrong size (should be $2==4, not $2==2). Both
patterns were redundant. They rewrote loc mlu as loc mli and loc slu
as loc sli, but this table doesn't have patterns for loc mli or loc
sli, so it is enough to rewrite mlu as mli and slu as sli.
Add the tokens IND_RL_B, IND_RL_H, IND_RL_H_S, IND_RL_D, along with
the rules to use them. These rules emit shorter code. For example,
loading a byte becomes lis, lbz instead of lis, addi, lbz.
While making this, I wrongly set IND_RL_D to size 4. Then ncg made
infinite recursion in codegen() and stackupto(), until it crashed by
stack overflow. I correctly set IND_RL_D to size 8, preventing the
crash.
Remove coercion from LABEL to REG. The coercion never happens because
I have stopped putting LABEL on the stack. Also remove LABEL from set
ANY_BHW. Retain the move from LABEL to REG because pat gto uses it.
Remove li32 instruction, unused after the switch to the hi16, ha16,
lo16 syntax.
Remove COMMENT(...) lines from most moves. In my opinion, they took
too much space, both in the table and in the assembly output. The
stacking rules and coercions keep their COMMENT(...) lines.
In test GPR, don't write to RSCRATCH.
Fold several coercions into a single coercion from ANY_BHW uses REG.
Use REG instead of GPR in stack patterns. REG and GPR act the same,
because every GPR on the stack is a REG, but I want to be clear that I
expect a REG, not r0.
In code rules, sort SUM_RC before SORT_RR, so I can add SUM_RL later.
Remove rules to optimize loc loc cii loc loc cii. If $2==$4, the
peephole optimizer can optimize it. If $2!=$4, then the EM program is
missing a conversion from size $2 to size $4.
Remove rules to store a SEX_B with sti 1 or a SEX_H with sti 2. These
rules would never get used, unless the EM program is missing a
conversion from size 4 to size 1 or 2.
Use it to generate code like
lis r12,ha16[__II0]
lis r11,ha16[_f]
lfs f1,lo16[_f](r11)
lfs f2,lo16[__II0](r12)
fadds f13,f2,f1
stfs f13,lo16[_f](r11)
Here ncg has allocated r11 for ha16[_f]. We use r11 in lfs and again
in stfs. Before this change, we needed an extra lis before stfs,
because ncg did not remember that ha16[_f] was in a register.
This example has a gap between ha16[__II0] and lo16[__II0], because
the lo16 is not in the next instruction. This requires my previous
commit 1bf58cf for RELOLIS. There is a gap because ncg emits the lis
as soon as I allocate it. The "lfs f2,lo16[__II0](r12)" happens in a
coercion from IND_RL_W to FSREG. The coercion allocates one FSREG but
may not allocate any other registers. So I must allocate r12 earlier.
I allocate r12 in pat lae, but this causes a gap.
A 4-byte load from a label yields a token IND_RL_W. This token emits
either lis/lwz or lis/lfs, if we want a general-purpose register or a
floating-point register.
Remove the GPRINDIRECT token, and use the IND_RC_* tokens as operands
to instructions. We no longer need to unpack an IND_RC_* token and
repack it as a GPRINDIRECT to use it in an instruction.
Allow storing IND_ALL_B and IND_ALL_H in register variables. Create a
set ANY_BHW for anything that we can store in a regvar.
Push register variables on the stack without using GPRE, by changing
stwu to accept LOCAL. Then ncg will replace the string ">>> BUG IN
LOCAL" with the register name. (I copied ">>> BUG IN LOCAL" from
mach/arm/ncg/table.)
Fix the rule for "pat lil inreg($1)>0" to yield a IND_RC_W token, not
a register. We might need to kill the token with "kills MEMORY".
Rename CONST_ALL to CONST_STACK, because it only includes constants on
the stack, and excludes CONST tokens. Instructions still don't allow
CONST_STACK operands, so we still need to repack each CONST_STACK as a
CONST to use it in an instruction.
Rename LABEL_OFFSET_HI to just LABEL_HI, and same for LABEL_HA and
LABEL_HO.
The new relocation type RELOLIS handles these instructions:
lis RT, ha16[expr] == addis RT, r0, ha16[expr]
lis RT, hi16[expr] == addis RT, r0, hi16[expr]
RELOLIS stores a 32-bit value in the program text. In this value, the
high bit is a ha16 flag, the next 5 bits are the target register RT,
and the low bits are a signed 26-bit offset. The linker replaces this
value with the lis instruction.
The old RELOPPC relocated a ha16/lo16 or hi16/lo16 pair. The new
RELOLIS relocates only a ha16 or hi16, so it is no longer necessary to
have a matching lo16 in the next instruction. The disadvantage is
that RELOLIS has only a signed 26-bit offset, not a 32-bit offset.
Switch the assembler to use RELOLIS for ha16 or hi16 and RELO2 for
lo16. The li32 instruction still uses the old RELOPPC relocation.
This is not the same as my RELOPPC change from my recent mail to
tack-devel (https://sourceforge.net/p/tack/mailman/message/35651528/).
This commit is on a different branch. Here I am throwing away my
RELOPPC change and instead trying RELOLIS.
We should not include a system header file here, because
mach/proto/as/comm2.y goes through cpp twice. The include can cause
problems like https://github.com/davidgiven/ack/issues/1
Remove this include #<stdbool.h> and leave a comment pointing to the
includes in comm0.h. Change the few instances of bool, false, true,
to int, 0, 1.
Type word_t is for encoding the machine instructions. It only needs
32 bits for PowerPC. It was long (which can have 32 or 64 bits), and
there was a second type quad (which was uint32_t). Switch word_t to
uint32_t and replace quad with word_t.
Also change valu_t and ADDR_T away from long.
They must use OP_RA_RS_RB_C instead of OP_RS_RA_RB_C. The code
generator often sets RS and RA to the same register, so swapping them
causes no change in many programs.
I also rename OP_RS_RA_UI_CC to OP_RA_RS_UI_CC, and OP_RS_RA_C to
OP_RA_RS_C, because they already swap RA and RS.
r0 is a special case and can't be used when adding a register to a
constant. The few remaining users of the scratch register don't do
that. I removed other usages of the scratch register in 7c64dab,
5b5f774, 19f0eb8, f64b7d8.
Also don't delete addis r0, r0, 0. These instructions are special
cases that set r0 to zero. If we delete them, then r0 keeps its old
value.
I caught this bug because osxppc protects the .text segment against
writing. (linuxppc doesn't protect it.) A program tried to set r0 to
the NULL pointer, but top deleted the instruction, so r0 kept an old
return address pointing into .text. Later the program checked that r0
wasn't NULL, tried to write to address r0, and crashed.
The rewritten code rules bring 3 new features:
1. The new rules compare a small constant with a register by
reversing the comparison and using `cmpwi` or `cmplwi`. The old
rules put the constant in a register.
2. The new rules emit shorter code to yield the test results,
without referencing the tables in mach/powerpc/ncg/tge.s.
3. The new rules use the extended `beq` and relatives, not the
basic `bc`, in the assembly output.
I delete the old tristate tokens and the old moves, because they
confused me. Some of the old moves weren't really moves. For
example, `move R3, C0` and then `move C0, R0` did not move r3 to r0.
I rename C0 to CR0.
This fixes the coercion from IND_ALL_D to FREG. The coercion had
never happened, because IND_ALL_D had 8 bytes but FREG had 4 bytes.
Instead, ncg always stacked the IND_ALL_D and unstacked a FREG. The
stacking rule uses f0, so the code did load f0 with the indirect
value, push f0 to stack, load f1 to stack, move stack pointer. Now
that FREG has 8 bytes, ncg does the coercion, and the code just loads
f1 with the indirect value.
Always use 'kills ALL' when reaching a label, because our registers
and tokens have the wrong values if the program jumps to this label
from somewhere else.
When falling through a label, if the top element is in r3, then
require that the rest of the stack is in the real STACK, not in
registers or tokens.
I'm doing this to be certain that the missing constraints are not
causing bugs. I did not find any such bug, perhaps because the labels
are usually near other instructions (like conditional branches and
function calls) that stack or kill tokens.
This is for fef 8 and fif 8. I changed .fef8 so it no longer kills
r7, but I don't want to update the list. We already use "kills ALL"
for most other calls to libem.
The new features are the hi16/lo16 and ha16/lo16 syntax for
relocations, and the extended mnemonics like "blr".
Use ha16/lo16 to load some double floats with 2 instructions (lis/lfd)
instead of 3 (lis/ori/lfd).
Use the extended names for branches, comparisons, and bit rotations,
so I can more easily read the code. The new names often encode the
same machine instructions as the old names, except in a few places
where I changed the instructions.
Stop using andi. when we don't need to set cr0. In inn.s, I change
andi. to extrwi to extract the same bits. In los.s and sts.s, I
change "andi. r3, r3, ~3" to "clrrwi r3, r3, 2". This avoids setting
cr0 and also stops clearing the high 16 bits of r3.
In csa.s, los.s, sts.s, I change some comparisons and right shifts
from signed to unsigned (cmplw, cmplwi, srwi), because the sizes are
unsigned. In inn.s, the right shift can be signed (sraw) or unsigned
(srw), but I use srw because we don't need the carry bit.
In fef8.s, I save an instruction by using rlwinm instead of addis/andc
to rlwinm to clear a field. The code no longer kills r7. In both
fef8.s and fif8.s, I remove the list of killed registers.
Also remove some whitespace from ends of lines.
Also make a few changes to basic mnemonics. Fix typo in name of the
basic "creqv". Add the basic "addc" and relatives, because it would
be odd to have the extended "subc" without "addc". Fix the basic
"rldicl", "rldicr", "rldic", "rldimi" to correctly encode the 6-bit MB
field. Fix "slw" and relatives to correctly swap their RA and RS
operands.
Add many, but not all, of the extended mnemonics from IBM's Power ISA
Version 2.06 Book I Appendix E. (I used 2.06, published 2009, just
because I already had the PDF of it.) This commit includes mnemonics
for branching, subtraction, traps, bit rotation, and a few others,
like "mflr" and "nop". The assembler now understands branches like
`beq cr7, label` and bit shifts like `slwi r7, r7, 2`. These encode
the same machine instructions as the basic "bc" and "rlwinm".
Some operands to basic names become optional. The assembler no longer
requires the level in "sc" or the branch hint in "bcctr" and "bclr";
they default to zero. Some extended names take an optional branch
hint or condition register.
Some extended names are still missing. I don't provide names with
static branch prediction, like "beq+" or "bge-", because the assembler
parses '+' and '-' as operators, not as part of an instruction name.
I also don't provide some names that 2.06 has for moving to or from
the condition register or some special purpose registers, names like
"mtcr" or "mfuamr".
This commit also deletes some unused tokens and one unused yacc rule.
I need this so I can add more %token lines to mach/powerpc/as/mach2.c
The assembler's tempfile encoded each token in a byte. This only
worked with tokens 0 to 127 and 256 and 383. If a token 384 or higher
existed, the assembler stopped working. I need tokens 384 and higher.
I change the token encoding to a 2-byte little-endian integer. I also
change a byte in the string encoding.
See issue #1 (https://github.com/davidgiven/ack/issues/1). The file
mach/proto/as/comm2.y goes through cpp twice. The _include macro,
defined in comm2.y and used in comm0.h, delays the inclusion of system
header files. The inclusion of <stdint.h> wasn't delayed. This
caused multiple inclusions of <sys/_types.h> in FreeBSD and
<machine/_types.h> in OpenBSD.
Use _include to delay <stdint.h>. Also use _include for "arch.h" and
"out.h", because h/out.h includes <stdint.h> and h/arch.h might
include it in the future.
Sort the system includes in comm0.h by moving them up to be with
<stdint.h>. Must include <stdint.h> before "mach0.c", because
mach/powerpc/as/mach0.c needs it. Must include "mach0.c" before
checking ASLD.
In my OpenBSD/amd64 system, the code becomes
if (0)
outname.on_valu &= ~(((0xFFFFFFFF)<<32)<<32);
The 0xFFFFFFFF is a 32-bit int, so the left shift by 32 is out of
range and causes the gcc warning.
The intent might be to clear any sign-extended bits, if the assignment
outname.on_valu = valu did sign extension. Old C had no unsigned
long, so .on_valu would have been long. The code is obsolete because
h/out.h now declares .on_valu as uint32_t.
possible values. Add the PowerPC ncg and mcg backend support to let the test
actually run, including modifying a bunch of PowrePC libem functions so that
they can be called from both ncg and mcg.
assembler directives, ha16() and has16(), for the upper half; has16() applies
the sign adjustment. .powerpcfixup is now gone, as we generate the relocation
in ha*() instead. Add special logic to the linker for undoing and redoing the
sign adjustment when reading/writing fixups. Tests still pass.
This provides and, ior, xor, com, zer, set, cms when defined($1) and
ior, set when !defined($1). I don't provide the other operations
!defined($1) because our Modula-2 compiler hasn't used them.
I wrote a Modula-2 example in
https://gist.github.com/kernigh/add79662bb3c63ffb7c46d01dc8ae788
Put a dummy comment in mach/powerpc/libem/build.lua so git checkout
will touch that file. Without the touch, the build system doesn't see
the new *.s files.
We only implement 'los 4', 'sts 4', 'cmi 4', 'cmu 4', not for sizes
other than 4. Add clause $1==4.
We only implement inn when defined($1).
The rule for aar needs 'kills ALL' because it kills many registers,
like other rules that call libem.
This allows 'move {CONST, $1}, R3' with a small enough $1 to emit one
instruction (addi) instead of two instructions (addis, ori). The
CONST token confusingly isn't in the CONST_ALL set.
The spec says, "ASS w: Adjust the stack pointer by w-byte integer".
The w argument "can either be given as argument or on top of the
stack." Therefore, 'ass 4' would pop the 4-byte integer from the
stack, but 'ass' would pop the size w from the stack, then pop the
w-byte integer.
PowerPC ncg wrongly implemented 'ass' as if it was 'ass 4'. Fix it to
accept only 'ass 4'.
These instructions would load or store the EM heap pointer. They
don't work. Programs must use brk() or sbrk() in libsys.
The last file to use 'lor 2' and 'str 2' was lang/pc/libpc/sav.e in
the Pascal library. Commit c084f9f deleted the file, so we no longer
need rules 'lor 2' or 'str 2' to build the ACK.
PowerPC has a few hundred special-purpose registers. The assembler
had only accepted the names "xer", "lr", "ctr". Most programs use
only those three SPRs. If I add more names, they would almost never
get used, and they might conflict with labels.
I want to use "mfspr r3, 0x3f0" and "mtspr 0x3f0, r3" in
plat/qemu/boot.s to access register hid0 from supervisor mode.
corresponding invocation in the ncg table so the same helpers can be used for
both mcg and ncg. Add a new IR opcode, FARJUMP, which jumps to a helper
function but saves volatile registers.
occasionally the first hop of a block would try to rearrange its registers (due
to evicted throughs), resulting in the phi moves copying values into the wrong
registers.
register 'type'; now use int/float/long/double throughout to identify
registers. Lots of register allocator tweaks and table bugfixes --- we now get
through the dreading Mathlib.mod!
This would have happened later, if f14 to f31 became regvar (like r13
to r31 are now). I am doing it now because ncg is too slow for rules
"with FREG FREG uses FREG". We use such rules for adf 8 and other EM
instructions that operate on 2 floats. Like my last commit cfbc537,
this commit speeds ncg by removing choices for register allocation.
ncg is too slow with this many registers. A stack pattern "with GPR
GPR GPR" or "with REG REG REG" takes too long to pick registers,
causing ncg 8 to take about 2 seconds on each sti 8. I introduce
REG_PAIR and there are only 4 such pairs.
For programs that use sti 8 (including C programs that copy 8-byte
structs), this speed hack improves the ncg run from several seconds to
almost instantaneous.
Also add a few COMMENT(...) lines in stacking rules.
This fixes the SIGILL (illegal instruction) in startrek when firing
phasers. The 32-bit processors in my PowerPC Mac and in QEMU don't
have fctid, a 64-bit instruction.
I got the idea from mach/proto/fp/fif8.c to extract the exponent,
clear some bits to get an integer, then subtract the integer from
the original value to get the fraction.
Adjust some of the loi rules (and associated moves) so we can identify
the tokens that must be in MEMORY.
With this commit, I can navigate the Enterprise even if I comment out
my work-around from e22c888.
Because li32 always loads a label into a GPR, it is sufficient to
coerce LABEL to REG, then use IND_RC_W or IND_RC_D for indirection
through the label.
Now that SUM_RC always has a signed 16-bit constant, it happens that
the various IND_RC_* tokens also have a signed 16-bit constant, so
we no longer need to touch the scratch register.
When loc (load constant) pushes a constant, it now checks the value of
the constant and pushes any of 7 tokens. These tokens allow stack
patterns to recognize 16-bit signed integers (CONST2), 16-bit unsigned
integers (UCONST2), multiples of 0x10000 (CONST_HZ), and other
interesting forms of constants.
Use the new constant tokens in the rules for adi, sbi, and, ior, xor.
Adjust a few other rules to understand the new tokens.
Require that SUM_RC has a signed 16-bit constant, and OR_RC and XOR_RC
each have an unsigned 16-bit constant. The moves from SUM_RC, OR_RC,
XOR_RC to GPR no longer touch the scratch register, because the
constant is not too big.
