Commit graph

4533 commits

Author SHA1 Message Date
Aleksandar Markovic 1c8614c303
target/mips: Clean up dsp_helper.c
Remove several minor checkpatch warnings and errors.

Backports commit f49ab2e1e6ca4f218cc970c937f91f9c69c95dd3 from qemu
2019-06-03 11:14:31 -04:00
Mateja Marjanovic 4b272cbe93
target/mips: Add emulation of MMI instruction PCPYUD
Add emulation of MMI instruction PCPYUD. The emulation is implemented
using TCG front end operations directly to achieve better performance.

Backports commit fd487f83ea92d790559813c5a0a719c30ca9ecde from qemu
2019-06-03 11:08:37 -04:00
Mateja Marjanovic c5e3fc601c
target/mips: Add emulation of MMI instruction PCPYLD
Add emulation of MMI instruction PCPYLD. The emulation is implemented
using TCG front end operations directly to achieve better performance.

Backports commit b87eef31f2f8047077d79c3180e9c8e762d2a50f from qemu
2019-06-03 11:05:50 -04:00
Mateja Marjanovic 7443387030
target/mips: Add emulation of MMI instruction PCPYH
Add emulation of MMI instruction PCPYH. The emulation is implemented
using TCG front end operations directly to achieve better performance.

Backports commit d3434d9f785ddaf40e0fd521ded400643ac4be09 from qemu
2019-06-03 11:03:07 -04:00
Jules Irenge 9c7f2f2e78
target/mips: realign comments to fix checkpatch warnings
Realign comments to fix warnings issued by checkpatc.pl tool
"WARNING: Block comments use a leading /* on a separate line"
within "target/mips/cpu.h" file.

Backports commit 9e72f33d854b0a817c0d2fe4bca693b76f0fe776 from qemu
2019-05-28 19:49:59 -04:00
Jules Irenge cf39970750
target/mips: add or remove space to fix checkpatch errors
Add or remove space to fix errors issued by checkpatch.pl tool
"ERROR: spaces required around that..."
"ERROR: space required after that..."
"ERROR: space required before the open parenthesis"
"ERROR: space required after that..."
"ERROR: space prohibited between function name and open parenthesis"
"ERROR: code indent should never use tabs"
"ERROR: line over 90 characters"
within "target/mips/cpu.h" file.

Backports commit 8ebf2e1a68408068c0bcd0d02a783fd12f6a9cb5 from qemu
2019-05-28 19:48:11 -04:00
Jakub Jermář 5b25eb80af
mips: Decide to map PAGE_EXEC in map_address
This commit addresses QEMU Bug #1825311:

mips_cpu_handle_mmu_fault renders all accessed pages executable

It allows finer-grained control over whether the accessed page should
be executable by moving the decision to the underlying map_address
function, which has more information for this.

As a result, pages that have the XI bit set in the TLB and are accessed
for read/write, don't suddenly end up being executable.

Fixes: https://bugs.launchpad.net/qemu/+bug/1825311
Fixes: 2fb58b73746e ('target-mips: add RI and XI fields to TLB entry')

Backports commit 7353113fa482e697a77575086a41f429a01f8dc0 from qemu
2019-05-28 19:44:28 -04:00
Mateja Marjanovic 9e8aed043e
target/mips: Refactor and fix INSERT.<B|H|W|D> instructions
The old version of the helper for the INSERT.<B|H|W|D> MSA instructions
has been replaced with four helpers that don't use switch, and change
the endianness of the given index, when executed on a big endian host.

Backports commit c1c9a10fb1f7a6782711817c167a2c20b000fc12 from qemu
2019-05-28 19:42:28 -04:00
Mateja Marjanovic d6a8d25015
target/mips: Refactor and fix COPY_U.<B|H|W> instructions
The old version of the helper for the COPY_U.<B|H|W> MSA instructions
has been replaced with four helpers that don't use switch, and change
the endianness of the given index, when executed on a big endian host.

Backports commit 41d288582782cf8d63241ecb6efa1e4160fe78f7 from qemu
2019-05-28 19:39:22 -04:00
Mateja Marjanovic 54a33d1db3
target/mips: Refactor and fix COPY_S.<B|H|W|D> instructions
The old version of the helper for the COPY_S.<B|H|W|D> MSA instructions
has been replaced with four helpers that don't use switch, and change
the endianness of the given index, when executed on a big endian host.

Backports commit 631c467461496dcf6d6a3e4c3d27a1433e96868e from qemu
2019-05-28 19:36:14 -04:00
Mateja Marjanovic 6dd651af3a
target/mips: Fix MSA instructions ST.<B|H|W|D> on big endian host
Fix the case when the host is a big endian machine, and change
the approach toward ST.<B|H|W|D> instruction helpers.

Backports commit 6decc572dcedbf298ae30f8213b39c8b842a595a from qemu
2019-05-28 19:29:27 -04:00
Mateja Marjanovic 3ee6295d7f
target/mips: Fix MSA instructions LD.<B|H|W|D> on big endian host
Fix the case when the host is a big endian machine, and change
the approach toward LD.<B|H|W|D> instruction helpers.

Backports commit 83be6b54123a8f3c529554139f1d1e43356edf8d from qemu
2019-05-28 19:27:05 -04:00
Mateja Marjanovic 1527b25428
target/mips: Make the results of MOD_<U|S>.<B|H|W|D> the same as on hardware
MSA instructions MOD_<U|S>.<B|H|W|D> when dividing by zero,
didn't return the same value when executed on a referent hardware
(FPGA MIPS 64 r6, little endian) and when executed on QEMU, which
is not a real bug, because the result when dividing by zero is
UNPREDICTABLE [1] (page 255, 256).

[1] MIPS Architecture for Programmers
Volume IV-j: The MIPS64 SIMD
Architecture Module, Revision 1.12

Backports commit cf122bf8d2732d5d8647901ebaea596668aaaa3a from qemu
2019-05-28 19:25:00 -04:00
Mateja Marjanovic d712d3f226
target/mips: Make the results of DIV_<U|S>.<B|H|W|D> the same as on hardware
MSA instructions DIV_<U|S>.<B|H|W|D> when dividing by zero,
didn't return the same value when executed on a referent hardware
(FPGA MIPS 64 r6, little endian) and when executed on QEMU, which
is not a real bug, because the result when dividing by zero is
UNPREDICTABLE [1] (page 141, 142).

[1] MIPS Architecture for Programmers
Volume IV-j: The MIPS64 SIMD
Architecture Module, Revision 1.12

Backports commit d2a40a5f6938f30f44b536e997e1e89bb62b971c from qemu
2019-05-28 19:24:26 -04:00
Jonathan Behrens 1d6acaa604
target/riscv: Only flush TLB if SATP.ASID changes
There is an analogous change for ARM here:
https://patchwork.kernel.org/patch/10649857

Backports commit 1e0d985fa9136a563168a3da66f3d17820404ee2 from qemu
2019-05-28 19:22:51 -04:00
Jonathan Behrens 7922aa54c0
target/riscv: More accurate handling of CSR
According to the spec, "All bits besides SSIP, USIP, and UEIP in the sip
register are read-only." Further, if an interrupt is not delegated to mode x,
then "the corresponding bits in xip [...] should appear to be hardwired to
zero. This patch implements both of those requirements.

Backports commit 087b051a51a0c2a5bc1e8d435a484a8896b4176b from qemu
2019-05-28 19:22:04 -04:00
Richard Henderson d1ad8bf44c
target/riscv: Add checks for several RVC reserved operands
C.ADDI16SP, C.LWSP, C.JR, C.ADDIW, C.LDSP all have reserved
operands that were not diagnosed.

Backports commit 4cc16b3b9282e04fab8e84d136540757e82af019 from qemu
2019-05-28 19:20:36 -04:00
Alistair Francis aca20201d4
target/riscv: Add the HGATP register masks
Backports commit e06431108b0b1ef6ca76398d2b0b792ea24ae6bc from qemu
2019-05-28 19:19:00 -04:00
Alistair Francis 294297b646
target/riscv: Add the HSTATUS register masks
Backports commit d28b15a4d3b1e000ec7bf9090fe870cbc5f1eb2c from qemu
2019-05-28 19:18:28 -04:00
Alistair Francis 2e6d11ee47
target/riscv: Add Hypervisor CSR macros
Add the 1.10.1 Hypervisor CSRs and remove the 1.9.1 spec versions.

Backports commit 71f09a5bb48d0c51b87e70158407ec2db4a9c6e2 from qemu
2019-05-28 19:17:54 -04:00
Alistair Francis 47e4e047bc
target/riscv: Allow setting mstatus virtulisation bits
Backports commit 1f0419cb0475eebdbefea67483e85287f3af07a7 from qemu
2019-05-28 19:17:18 -04:00
Alistair Francis c64f57c360
target/riscv: Add the MPV and MTL mstatus bits
Backports commit 49aaa3e534f5422a56313bb93c1880e70fc1da7e from qemu
2019-05-28 19:15:33 -04:00
Alistair Francis b44de569f0
target/riscv: Improve the scause logic
No functional change, just making the code easier to read.

Backports commit 16fdb8ff64374ed51b246437e13043039a8eb9f9 from qemu
2019-05-28 19:14:44 -04:00
Alistair Francis 4b0355dcfc
target/riscv: Mark privilege level 2 as reserved
Backports commit 356d74192a035c71a78a22d24812a6df6099ae40 from qemu
2019-05-28 19:12:10 -04:00
Alistair Francis ea2fee2d4d
target/riscv: Add a base 32 and 64 bit CPU
At the same time deprecate the ISA string CPUs.

It is dobtful anyone specifies the CPUs, but we are keeping them for the
Spike machine (which is about to be depreated) so we may as well just
mark them as deprecated.

Backports commit 8903bf6e6d73d03b988b4a8197132de2ad681ff5 from qemu
2019-05-28 19:11:12 -04:00
Richard Henderson 9c1212f627
target/riscv: Remove spaces from register names
These extra spaces make the "-d op" dump look weird.

Backports commit 7f9188e210aff6522a960d9669a583a3a752ddc0 from qemu
2019-05-28 19:08:50 -04:00
Richard Henderson 68ce00ac2f
target/riscv: Split gen_arith_imm into functional and temp
The tcg_gen_fooi_tl functions have some immediate constant
folding built in, which match up with some of the riscv asm
builtin macros, like mv and not.

Backports commit 598aa1160c3d17ab9271daf1f69d093ebada3f25 from qemu
2019-05-28 19:07:53 -04:00
Richard Henderson a62b4e5def
target/riscv: Split RVC32 and RVC64 insns into separate files
This eliminates all functions in insn_trans/trans_rvc.inc.c,
so the entire file can be removed.

Backports commit 0e68e240a9bd3b44a91cd6012f0e2bf2a43b9fe2 from qemu
2019-05-28 19:00:23 -04:00
Richard Henderson a968769d26
target/riscv: Use pattern groups in insn16.decode
This eliminates about half of the complicated decode
bits within insn_trans/trans_rvc.inc.c.

Backports commit c2cfb97c01a3636867c1a4a24f8a99fd8c6bed28 from qemu
2019-05-28 18:55:28 -04:00
Richard Henderson 8360c1fa3b
target/riscv: Merge argument decode for RVC shifti
Special handling for IMM==0 is the only difference between
RVC shifti and RVI shifti. This can be handled with !function.

Backports commit 6cafec92f1c862a9754ef6a28be68ba7178a284d from qemu
2019-05-28 18:52:50 -04:00
Richard Henderson dc087c4c0c
target/riscv: Merge argument sets for insn32 and insn16
In some cases this allows us to directly use the insn32
translator function. In some cases we still need a shim.

Backports commit e1d455dd91c935c714412dafeb24db947429a929 from qemu
2019-05-28 18:50:48 -04:00
Richard Henderson cb2ce66814
target/riscv: Use --static-decode for decodetree
The generated functions are only used within translate.c
and do not need to be global, or declared.

Backports commit 81770255581bd210c57b86a6e808628ab8d0c543 from qemu
2019-05-28 18:45:23 -04:00
Richard Henderson d51505f6e9
target/riscv: Name the argument sets for all of insn32 formats
Backports commit e761799796ac2211b9706753c459e117e7be58fa from qemu
2019-05-28 18:36:53 -04:00
Fabien Chouteau 7e6d37b51d
RISC-V: fix single stepping over ret and other branching instructions
This patch introduces wrappers around the tcg_gen_exit_tb() and
tcg_gen_lookup_and_goto_ptr() functions that handle single stepping,
i.e. call gen_exception_debug() when single stepping is enabled.

Theses functions are then used instead of the originals, bringing single
stepping handling in places where it was previously ignored such as jalr
and system branch instructions (ecall, mret, sret, etc.).

Backports commit 6e2716d8ca4edf3597307accef7af36e8ad966eb from qemu
2019-05-28 18:35:07 -04:00
Jonathan Behrens 25c0333213
target/riscv: Do not allow sfence.vma from user mode
The 'sfence.vma' instruction is privileged, and should only ever be allowed
when executing in supervisor mode or higher.

Backports commit b86f4167630802128d94f3c89043d97d2f4c2546 from qemu
2019-05-28 18:29:46 -04:00
Richard Henderson 67f0af4282
tcg/aarch64: Allow immediates for vector ORR and BIC
The allows immediates to be used for ORR and BIC,
as well as the trivial inversions, ORC and AND.

Backports commit 9e27f58b9902834dffc0d66d9eb62f78d9c2a632 from qemu
2019-05-24 18:47:07 -04:00
Richard Henderson 5ecfba4fe6
tcg/aarch64: Build vector immediates with two insns
Use MOVI+ORR or MVNI+BIC in order to build some vector constants,
as opposed to dropping them to the constant pool. This includes
all 16-bit constants and a similar set of 32-bit constants.

Backports commit 02f3a5b4744885258758d07ebe09cf965de78bcf from qemu
2019-05-24 18:43:54 -04:00
Richard Henderson 06058ef648
tcg/aarch64: Use MVNI in tcg_out_dupi_vec
The compliment of a subset of immediates can be computed
with a single instruction.

Backports commit 7e308e003e5b6ddd3130e09711e1d33693230696 from qemu
2019-05-24 18:42:40 -04:00
Richard Henderson c18ec586dc
tcg/aarch64: Split up is_fimm
There are several sub-classes of vector immediate, and only MOVI
can use them all. This will enable usage of MVNI and ORRI, which
use progressively fewer sub-classes.

This patch adds no new functionality, merely splits the function
and moves part of the logic into tcg_out_dupi_vec.

Backports commit 984fdcee342473dfe797897758929dad654693c8 from qemu
2019-05-24 18:41:37 -04:00
Richard Henderson 0ea4c05dc3
tcg/aarch64: Support vector bitwise select value
The instruction set has 3 insns that perform the same operation,
only varying in which operand must overlap the destination. We
can represent the operation without overlap and choose based on
the operands seen.

Backports commit a9e434a5dc16f71ee156428619fc3c3765b68f26 from qemu
2019-05-24 18:38:37 -04:00
Richard Henderson c79510378f
tcg/i386: Use umin/umax in expanding unsigned compare
Using umin(a, b) == a as an expansion for TCG_COND_LEU is a
better alternative to (a - INT_MIN) <= (b - INT_MIN).

Backports commit ebcfb91abed8c0fb180a968b9004419c208dcc02 from qemu
2019-05-24 18:36:32 -04:00
Richard Henderson ffdbc1a233
tcg/i386: Remove expansion for missing minmax
This is now handled by code within tcg-op-vec.c.

Backports commit 3ec3538a45f2fead475b0cca6945092c87927b4f from qemu
2019-05-24 18:34:44 -04:00
Richard Henderson 68cb096196
tcg/i386: Support vector comparison select value
We already had backend support for this feature. Expand the new
cmpsel opcode using vpblendb. The combination allows us to avoid
an extra NOT for some comparison codes.

Backports commit 904c5e19672778cc3349f4975437cfdf3371abb6 from qemu
2019-05-24 18:33:16 -04:00
Richard Henderson a868533297
tcg: Add TCG_OPF_NOT_PRESENT if TCG_TARGET_HAS_foo is negative
If INDEX_op_foo is always expanded by tcg_expand_vec_op, then
there may be no reasonable set of constraints to return from
tcg_target_op_def for that opcode.

Let TCG_TARGET_HAS_foo be specified as -1 in that case. Thus a
boolean test for TCG_TARGET_HAS_foo is true, but we will not
assert within process_op_defs when no constraints are specified.

Compare this with tcg_can_emit_vec_op, which already uses this
tri-state indication.

Backports commit 25c012b4009256505be3430480954a0233de343e from qemu
2019-05-24 18:28:11 -04:00
Richard Henderson 568da655c6
tcg: Expand vector minmax using cmp+cmpsel
Provide a generic fallback for the min/max operations.

Backports commit 72b4c792c7a576d9246207a8e9a940ed9e191722 from qemu
2019-05-24 18:26:53 -04:00
Richard Henderson 56d35e80aa
tcg: Introduce do_op3_nofail for vector expansion
This makes do_op3 match do_op2 in allowing for failure,
and thus fall back expansions.

Backports commit 17f79944ebeace8bf43047a33b7775ba5ed9070e from qemu
2019-05-24 18:24:44 -04:00
Richard Henderson 2ea6dfbd63
tcg: Add support for vector compare select
Perform a per-element conditional move. This combination operation is
easier to implement on some host vector units than plain cmp+bitsel.
Omit the usual gvec interface, as this is intended to be used by
target-specific gvec expansion call-backs.

Backports commit f75da2988eb2457fa23d006d573220c5c680ec4e from qemu
2019-05-24 18:21:13 -04:00
Richard Henderson ca58be9cb4
tcg: Add support for vector bitwise select
This operation performs d = (b & a) | (c & ~a), and is present
on a majority of host vector units. Include gvec expanders.

Backports commit 38dc12947ec9106237f9cdbd428792c985cd86ae from qemu
2019-05-24 18:15:10 -04:00
Richard Henderson fa363c3d6d
tcg: Fix missing checks and clears in tcg_gen_gvec_dup_mem
The paths through tcg_gen_dup_mem_vec and through MO_128 were
missing the check_size_align. The path through MO_128 was also
missing the expand_clr. This last was not visible because the
only user is ARM SVE, which would set oprsz == maxsz, and not
require the clear.

Fix by adding the check_size_align and using do_dup directly
instead of duplicating the check in tcg_gen_gvec_dup_{i32,i64}.

Backports commit 532ba368a13712724137228b5e7e9435994d25e1 from qemu
2019-05-24 18:07:28 -04:00
Richard Henderson 60cfe541b2
tcg/i386: Fix dupi/dupm for avx1 and 32-bit hosts
The VBROADCASTSD instruction only allows %ymm registers as destination.
Rather than forcing VEX.L and writing to the entire 256-bit register,
revert to using MOVDDUP with an %xmm register. This is sufficient for
an avx1 host since we do not support TCG_TYPE_V256 for that case.

Also fix the 32-bit avx2, which should have used VPBROADCASTW.

Fixes: 1e262b49b533

Backports commit 7b60ef3264e9627ac6efb34e9a6130647e9b55c0 from qemu
2019-05-24 18:04:08 -04:00