The beginning of the register allocation order list on the TCG arm
target matches the list of clobbered registers. This means that when an
helper is called, there is almost always clobbered registers that have
to be spilled.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
64-bit arguments should be aligned on an even register as specified
by the "Procedure Call Standard for the ARM Architecture".
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
addr_reg, data_reg and data_reg2 can't be register r0 or r1 du to the
constraints. Don't check if they equals these registers.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
On big endian targets, data arguments of qemu_ld/st ops have to be
byte swapped. Two temporary registers are needed for qemu_st to do
the bswap. r0 and r1 are used in system mode, do the same in user
mode, which implies reworking the constraints.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
While it make sense to pass a conditional argument to tcg_out_*()
functions as the ARM architecture allows that, it doesn't make sense
for qemu_ld/st functions. These functions use comparison instructions
and conditional execution already, so it is not possible to use a
second level of conditional execution.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Add an bswap16 and bswap32 ops, either using the rev and rev16
instructions on ARMv6+ or shifts and logical operations on previous
ARM versions. In both cases the result use less instructions than
the pure TCG version.
These ops are also needed by the qemu_ld/st functions.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Add an ext16u op, either using the uxth instruction on ARMv6+ or two
shifts on previous ARM versions. In both cases the result use the same
number or less instructions than the pure TCG version.
Also move all sign extension code to separate functions, so that they
can be reused in other parts of the code.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Use a set of variables to define the allowed ARM instructions, depending
on the __ARM_ARCH_*__ GCC defines.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The TCG ARM backends uses integer values to refer to both immediate
values and register number. This makes the code difficult to read.
The patch below replaces all (if I haven't miss any ;-) integer values
representing register number by TCG_REG_* enum values.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Instead of writing very compact code, declare all registers that are
clobbered or reserved one by one. This makes the code easier to read.
Also declare all the 16 registers to TCG, and mark pc as reserved.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
There is no need to save the LR register (r14) before a call to a
subroutine. According to the "Procedure Call Standard for the ARM
Architecture", it is the job of the callee to save this register.
Moreover, this register is already saved in the prologue/epilogue.
This patch removes the disabled SAVE_LR code, as there is no need to
reenable later.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
There's a header file inclusion ordering problem between cpu-all.h
and qemu-timer.h, such that cpu_get_real_ticks is not defined when
we attempt to use it in profile_getclock.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Register fp (frame pointer) is a bad choice for compilations
without optimisation, because the compiler makes heavy use
of this register (so the resulting code crashes).
Register s0 had been used for TCG_AREG1 in earlier releases,
but was no longer used and is now free for TCG_AREG0.
The resulting code works for compilations without
optimisation (tested with qemu mips in qemu mips
on x86 host).
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
PA-RISC uses procedure descriptors. We'd need to emit a call to
the millicode routine $$dyncall. However, this situation doesn't
actually arise, since we always have the descriptor available at
TCG code generation time.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Handle the output log part overlapping the input high parts.
Also, improve sub2 to handle some constants the second input low part.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Delete inline functions from tcg-target.h that don't need to be there,
move the others to tcg-target.c. Add 'Z', 'I', 'J' constraints for
0, signed 11-bit, and signed 5-bit respectively. Add GUEST_BASE support
similar to ppc64, with the value stored in a register. Add missing
registers to reg_alloc_order. Add support for 12-bit branch relocations.
Add functions for synthetic operations: addi, mtctl, dep, shd, vshd, ori,
andi, shifts, rotates, multiply, branches, setcond. Split out TLB reads
from qemu_ld and qemu_st; fix argument loading for tlb external calls.
Generate the prologue.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Historically the qemu tlb "addend" field was used for both RAM and IO accesses,
so needed to be able to hold both host addresses (unsigned long) and guest
physical addresses (target_phys_addr_t). However since the introduction of
the iotlb field it has only been used for RAM accesses.
This means we can change the type of addend to unsigned long, and remove
associated hacks in the big-endian TCG backends.
We can also remove the host dependence from target_phys_addr_t.
Signed-off-by: Paul Brook <paul@codesourcery.com>
A few words about design choices:
* On IA64, instructions should be grouped by bundle, and dependencies
between instructions declared. A first version of this code tried to
schedule instructions automatically, but was very complex and too
invasive for the current common TCG code (ops not ending at
instruction boundaries, code retranslation breaking already generated
code, etc.) It was also not very efficient, as dependencies between
TCG ops is not available.
Instead the option taken by the current implementation does not try
to fill the bundle by scheduling instructions, but by providing ops
not available as an ia64 instruction, and by offering 22-bit constant
loading for most of the instructions. With both options the bundle are
filled at approximately the same level.
* Up to 128 registers can be affected to a function on IA64, but TCG
limits this number to 64, which is actually more than enough. The
register affectation is the following:
- r0: used to map a constant argument with value 0
- r1: global pointer
- r2, r3: internal use
- r4 to r6: not used to avoid saving them
- r7: env structure
- r8 to r11: free for TCG (call clobbered)
- r12: stack pointer
- r13: thread pointer
- r14 to r31: free for TCG (call clobbered)
- r32: reserved (return address)
- r33: reserved (PFS)
- r33 to r63: free for TCG
* The IA64 architecture has only 64-bit registers and no 32-bit
instructions (the only exception being cmp4). Therefore 64-bit
registers and instructions are used for 32-bit ops. The adopted
strategy is the same as the ABI, that is the higher 32 bits are
undefined. Most ops (and, or, add, shl, etc.) can directly use
the 64-bit registers, while some others have to sign-extend (sar,
div, etc.) or zero-extend (shr, divu, etc.) the register first.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Commit 86feb1c860
did not change all occurrences of INDEX_op_qemu_ld32u
for tcg/arm.
Please note that I could not test this patch
(I have currently no arm system available).
Cc: Richard Henderson <rth@twiddle.net>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Some targets (e.g. Alpha and MIPS64) need to keep 32-bit operands
sign-extended in 64-bit registers (regardless of the "real" sign
of the operand). For that, we need to be able to distinguish
between a 32-bit load with a 32-bit result and a 32-bit load with
a given extension to a 64-bit result. This distinction already
exists for the ld* loads, but not the qemu_ld* loads.
Reserve qemu_ld32u for 64-bit outputs and introduce qemu_ld32 for
32-bit outputs. Adjust all code generators to match.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The TCGType name was already used consistently. Changing it
to an enumeration instead of a set of defines aids debugging.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Use the TCGCond enumeration type in the brcond and setcond
related prototypes in tcg-op.h and each code generator.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Give the enumeration formed from tcg-opc.h a name: TCGOpcode.
Use that enumeration type instead of "int" whereever appropriate.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
There is no need to save r7, it is used to store the address
of the env structure and is not modified by GCC.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
TCG internal helpers only access to the values passed in arguments, and
do not modify the CPU internal state. Thus they can be declared as
const and pure.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Some targets like ARM would benefit to use 32-bit helpers for
div/rem/divu/remu.
Create a #define for div2 so that targets can select between
div, div2 and helper implementation. Use the helper version if none
of the #define are present.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Since commit 6113d6d316 QEMU crashes
on ARM hosts. This is not a bug of this commit, but a latent bug
revealed by this commit.
The TCG code is called through a procedure call using the prologue
and epilogue code. This code does not save and restore enough registers.
The "Procedure Call Standard for the ARM Architecture" says:
A subroutine must preserve the contents of the registers r4-r8, r10,
r11 and SP (and r9 in PCS variants that designate r9 as v6).
The current code only saves and restores r9 to r11, and misses r4 to
r8. The patch fixes that by saving r4 to r12. Theoretically there is
no need to save and restore r12, but an even number of registers have
to be saved as per EABI.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Fix error:
CC sparc-bsd-user/op_helper.o
In file included from /src/qemu/tcg/tcg.c:158:
/src/qemu/tcg/sparc/tcg-target.c:728:5: "TARGET_PHYS_ADDR_BITS" is not defined
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
When restoring register values, increase the stack register for skipped
values.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Andrzej Zaborowski <andrew.zaborowski@intel.com>
On 32-bit hosts op_qemu_ld32s is unused. Remove it to fix the
following assertion failure:
qemu-alpha: tcg/tcg.c:1055:
tcg_add_target_add_op_defs: Assertion `tcg_op_defs[op].used' failed.
Signed-off-by: Jay Foad <jay.foad@gmail.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Previously ORC was always implemented by tcg-op.h with
an explicit NOT opcode. Allow a target implementation.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Previously ANDC was always implemented by tcg-op.h with
an explicit NOT opcode. Allow a target implementation.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
The fallback implementation of "ret = arg1 ^ -1" isn't ideal
because of the extra tcg op to load the minus one.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
The fallback implementation of "ret = 0 - arg1" isn't ideal,
first because of the extra tcg op to load the zero, and second
because we fail to handle zero as %g0 for arg1 of the sub.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
New version after malc's comments. (This avoids having to do
#if defined __linux__ || defined __FreeBSD__ || defined __FreeBSD_kernel__
for the third case.)
Submitted by: Andreas Tobler <andreast@fgznet.ch> (original version)
Signed-off-by: Juergen Lock <nox@jelal.kn-bremen.de>
Signed-off-by: malc <av1474@comtv.ru>
When compiled with CONFIG_DEBUG_TCG, this code looks
for missing, duplicate and wrong entries in the
op definitions.
Errors will raise an assertion at program start
(all checks are done in the initial phase).
The current code contains such errors, at least for
i386 guest on i386 host.
Signed-off-by: Stefan Weil <weil@mail.berlios.de>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Returns the condition as if with swapped comparison operands.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
The address register is overriden when it corresponds to v0 and the fast
path is taken, which leads to a crash. Fix that by using the a0 register
instead.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
There are places, like brcond2, where we know that the destination
of a forward branch will be within 127 bytes.
Add the R_386_PC8 relocation type to support this. Add a flag to
tcg_out_jxx and tcg_out_brcond* to enable it. Set the flag in the
brcond2 label_next branches; pass along the input flag otherwise.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
It is very handy to have a reliable mapping of a condition to its inverse.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Defines setcond_{i32,i64} and setcond2_i32 for 64-on-32-bit.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The internal opcodes brcond2, add2, sub2, mulu2 were undocumented.
Place these in a new section that clearly indicates that they are
not to be emitted by translators.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The existing P_REXB internal opcode flag unconditionally emits
the REX prefix. Technically it's not needed if the register in
question is %al, %bl, %cl, %dl.
Eliding the prefix requires splitting the P_REXB flag into two,
in order to indicate whether the byte register in question is
in the REG or the R/M field. Within TCG, the byte register is
in the REG field only for stores.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
This avoids an unnecessary REX.W prefix when dealing with AND
operands that fit into a 32-bit quantity. The most common change
actually seen is movz[wb]q -> movz[wb]l.
Similarly, avoid REXW in ext{8,16}u_i64 tcg opcodes.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The 32-bit right-shift instructions is defined to extend the shifted
output to 64-bits. A shift count of zero therefore is a simple
extension without actually shifting.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
The {div,divu}2 opcodes are intended for systems for which the
division instruction produces both quotient and remainder. Sparc
is not such a system. Indeed, the remainder must be computed as
quot = a / b
rem = a - (quot * b)
Split out a tcg_out_div32 function that properly initializes Y
with the extension of the input to 64-bits. Discard the code
that used the 64-bit DIVX on sparc9/sparcv8plus without extending
the inputs to 64-bits. Implement remainders in terms of division
followed by multiplication.
Signed-off-by: Richard Henderson <rth@twiddle.net>
[blauwirbel@gmail.com: applied rth's typo fix in tcg_out_div32]
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Add a function to handle the register-vs-immediate test for arithmetic.
Also, adjust the OP_32_64 macro so that it auto-indents properly.
Rename the gen_arith32 label to gen_arith, since it handles 64-bit
arithmetic as well.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Returns an unsigned version of a signed condition;
returns the original condition otherwise.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Split out tcg_out_cmp and properly handle immediate arguments.
Fix constraints on brcond to match what SUBCC accepts.
Add tcg_out_brcond2_i32 for 32-bit host.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
The test TCG_TARGET_REG_BITS==64 is exactly the feature that we
are checking for, whereas something involving __sparc_v9__ or
__sparc_v8plus__ should be reserved for something ISA related,
as with SMULX.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Generate sign-extended 32-bit constants with SETHI+XOR.
Otherwise tidy the routine to avoid the need for
conditional compilation and code duplication with movi_imm32.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
We were unnecessarily restricting imm13 constants to 12 bits.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Qemu won't let us run a KVM target without having host TCG support. Well, for
now we don't have any so let's implement a fake target that only stubs out
everything.
I tried to keep the patch as close to Uli's source as possible, so whenever
he feels like it he can easily diff his version against this one.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Based on a patch from Arnaud Patard (Rtp) <arnaud.patard@rtp-net.org>
A few words about design choices:
* Two registers, at and t0, are reserved for TCG internal use. They are
useful for bswap and 64-bit ops.
* Most ops supports a constant argument with value 0, which is actually
mapped to the zero register.
* While the at register is available for constant loading, ops only
support a limited range of constants. TCG does a better job doing the
register allocation and constant loading by itself. There are plenty of
registers available anyway.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Similarly to what is already done in tcg_liveness_analysis() when
USE_LIVENESS_ANALYSIS is not set.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Currently zero extensions ops are implemented by a and op with a
constant. This is then catched in some backend, and replaced by
a zero extension instruction. While this works well on RISC
machines, this adds a useless register move on non-RISC machines.
Example on x86:
ext16u_i32 r1, r2
is translated into
mov %eax,%ebx
movzwl %bx, %ebx
while the optimized version should be:
movzwl %ax, %ebx
This patch adds ext{8,16,32}u_i{32,64} TCG ops that can be
implemented in the backends to avoid emitting useless register
moves.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
This patch uses sxtb for ext8s_i32 and sxth for ext16s_i32 in ARM back-end.
Signed-off-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
the encode_imm function in tcg/arm/tcg-target.c lacks shift declaration.
Laurent
Signed-off-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Andrzej Zaborowski <andrew.zaborowski@intel.com>
this patch handles all possible constants for immediate operand of ALU ops.
I'm not very satisfied by the implementation.
Laurent
Signed-off-by: Laurent Desnogues <laurent.desnogues@gmail.com>
Signed-off-by: Andrzej Zaborowski <andrew.zaborowski@intel.com>
