The TriCore documentation was wrong on how to calculate ovf bits for those two
instructions, which I confirmed with real hardware (TC1796 chip). An ovf
actually happens, if the result (without remainder) does not fit into 8/16 bits.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
RRPW_DEXTR used r1 for the low part and r2 for the high part. It should be the
other way round. This also fixes that the result of the first shift was not
saved in a temp and could overwrite registers that were needed for the second
shift.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
dvinit_hu/bu for ISA v1.3 calculate the higher part of the result, that is needed
for the overflow bits, after calculating the overflow bits.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
clang report:
target-tricore/op_helper.c:1247:24: warning:
taking the absolute value of unsigned type 'uint32_t' (aka 'unsigned int')
has no effect [-Wabsolute-value]
target-tricore/op_helper.c:1248:25: warning:
taking the absolute value of unsigned type 'uint32_t' (aka 'unsigned int')
has no effect [-Wabsolute-value]
target-tricore/op_helper.c:1249:19: warning:
taking the absolute value of unsigned type 'uint32_t' (aka 'unsigned int')
has no effect [-Wabsolute-value]
target-tricore/op_helper.c:1297:24: warning:
taking the absolute value of unsigned type 'uint32_t' (aka 'unsigned int')
has no effect [-Wabsolute-value]
target-tricore/op_helper.c:1298:25: warning:
taking the absolute value of unsigned type 'uint32_t' (aka 'unsigned int')
has no effect [-Wabsolute-value]
target-tricore/op_helper.c:1299:19: warning:
taking the absolute value of unsigned type 'uint32_t' (aka 'unsigned int')
has no effect [-Wabsolute-value]
Fix also the divisor which was taken from the wrong register
(thanks to Peter Maydell for this hint).
Cc: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Message-Id: <1425739412-8144-1-git-send-email-sw@weilnetz.de>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Add helpers helper_subadr_h/_ssov which subs one halfword and adds one
halfword, rounds / and saturates each half word independently.
Add microcode helper functions:
* gen_msubad_h/ads_h: multiply two halfwords left justified and sub from the
first one word and add the second one word
/ and saturate each resulting word independetly.
* gen_msubadm_h/adms_h: multiply two halfwords in q-format left justified
and sub from the first one word and add to
the second one word / and saturate each resulting
word independetly.
* gen_msubadr32_h/32s_h: multiply two halfwords in q-format left justified
and sub from the first one word and add to
the second one word, round both results / and
saturate each resulting word independetly.
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Add helpers:
* msub64_q_ssov: multiply two 32 bit q-format number, sub the result from a
64 bit q-format number and saturate.
* msub32_q_sub_ssov: sub two 64 bit q-format numbers and return a 32 bit
result.
* msubr_q_ssov: multiply two 32 bit q-format numbers, sub the result from a 32 bit
q-format number and saturate.
* msubr_q: multiply two 32 bit q-format numbers and sub the result from a 32 bit
q-format number.
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Add helpers:
* sub64_ssov: subs two 64 bit values and saturates the result.
* subr_h/_ssov: subs two halfwords from two words in q-format with rounding
/ and saturates each result independetly.
Add microcode generator:
* gen_sub64_d: adds two 64 bit values.
* gen_msub_h/s_h: multiply four halfwords, sub each result left justfied
from two word values / and saturate each result.
* gen_msubm_h/s_h: multiply four halfwords, sub each result left justfied
from two words values in q-format / and saturate each
result.
* gen_msubr32/64_h/s_h: multiply four halfwords, sub each result left
justfied from two halftwords/words values in q-format
/ and saturate each result.
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
This is improved type checking for the translators -- it's no longer
possible to accidentally swap arguments to the branch functions.
Note that the code generating backends still manipulate labels as int.
With notable exceptions, the scope of the change is just a few lines
for each target, so it's not worth building extra machinery to do this
change in per-target increments.
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Edgar E. Iglesias <edgar.iglesias@gmail.com>
Cc: Michael Walle <michael@walle.cc>
Cc: Leon Alrae <leon.alrae@imgtec.com>
Cc: Anthony Green <green@moxielogic.com>
Cc: Jia Liu <proljc@gmail.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: Aurelien Jarno <aurelien@aurel32.net>
Cc: Blue Swirl <blauwirbel@gmail.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Add helpers helper_addsur_h/_ssov which adds one halfword and subtracts one
halfword, rounds / and saturates each half word independently.
Add microcode helper functions:
* gen_maddsu_h/sus_h: multiply two halfwords left justified and add to the
first one word and subtract from the second one word
/ and saturate each resulting word independetly.
* gen_maddsum_h/sums_h: multiply two halfwords in q-format left justified
and add to the first one word and subtract from
the second one word / and saturate each resulting
word independetly.
* gen_maddsur32_h/32s_h: multiply two halfwords in q-format left justified
and add to the first one word and subtract from
the second one word, round both results / and
saturate each resulting word independetly.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add helpers:
* madd64_q_ssov: multiply two 32 bit q-format number, add them with a
64 bit q-format number and saturate.
* madd32_q_add_ssov: add two 64 bit q-format numbers and return a 32 bit
result.
* maddr_q_ssov: multiplay two 32 bit q-format numbers, add a 32 bit
q-format number and saturate.
* maddr_q: multiplay two 32 bit q-format numbers and add a 32 bit
q-format number.
Note: madd instructions in the q format can behave strange, e.g.
0x1 + (0x80000000 * 0x80000000) << 1 for 32 bit signed values does not cause an
overflow on the guest, because all intermediate results should be handled as if
they are indefinitely precise. We handle this by inverting the overflow bit for
all cases: a + (0x80000000 * 0x80000000) << 1.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add helpers:
* add64_ssov: adds two 64 bit values and saturates the result.
* addr_h/_ssov: adds two halfwords with two words in q-format with rounding
/ and saturates each result independetly.
Add microcode generator:
* gen_add64_d: adds two 64 bit values.
* gen_addsub64_h: adds/subtracts one halfwords with a word and adds/
subtracts another halftword with another word.
* gen_madd_h/s_h: multiply four halfwords, add each result left justfied
to two word values / and saturate each result.
* gen_maddm_h/s_h: multiply four halfwords, add each result left justfied
to two words values in q-format / and saturate each
result.
* gen_maddr32/64_h/s_h: multiply four halfwords, add each result left
justfied to two halftwords/words values in q-format
/ and saturate each result.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
If the signed result of the multiplication overflows, we would get a negative
value, which would result in a addition instead of a subtraction.
Now we do the overflow calculation and saturation by hand instead of using
suov32_neg.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
The method by which we count the number of ops emitted
is going to change. Abstract that away into some inlines.
Reviewed-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Add microcode generator function gen_cond_sub.
Add helper functions:
* ixmax/ixmin: search for the max/min value and its related index in a
vector of 16-bit values.
* pack: dack two data registers into an IEEE-754 single precision floating
point format number.
* dvadj: divide-adjust the result after dvstep instructions.
* dvstep: divide a reg by a divisor, producing 8-bits of quotient at a time.
OPCM_32_RRR_FLOAT -> OPCM_32_RRR_DIVIDE
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
suov checks unsigned for an overflow and an underflow, after some arithmetic
operations and saturates the result to either max_uint32 or 0. So far we
handled this by expanding to the next bigger data type and compare whether
the result is > max_uint32 or < 0.
However this approach can fail for an 32 bit multiplication, if both operands of
the multiplication are 0x80000000. This sets the sign bit of the 64 bit integer
and would result in a false saturation to 0.
Since unsigned operations, e.g add, sub, mul always result in either a positive
or negative overflow, we split the functions for suov32 up into two functions
(suov32_pos, suov32_neg) for each case.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
This fixes one bug and one false positive found by coverity. The bug is,
that gen_mtcr was missing a mask to check the flag, which resulted in dead code.
The false positive is a intentional missing break for a jump and link address
insn followed by a jump and link insn. This adds a fall through comment to avoid
the false positive in the future.
Reported-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
64 bit mac instructions calculated the av bits after the saturation, which
resulted in a wrong PSW. This moves the av bit calculation before the
saturation.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Fix tc1796 cpu model using wrong ISA version.
Fix cond_add sometimes writing back wrong result.
Fix RCR_SEL and RCR_SELN using wrong registers for result and cond.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Add a missing ULL suffix to a 64 bit constant: this suppresses a
compiler warning from mingw32 gcc.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
the sdhci-pci device id to make space for the rocker device.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)
iQEcBAABAgAGBQJUp6bMAAoJEL/70l94x66D9q0H/06JFtq/Sn5jLJsWgkU2EcNM
VpV9E2d+RE711PuKkplTj1ee/oYLTnxG9Yf+gdR3KY0hjB/32tHi70ft1WyvbGh/
m/9AiRMmVsKqnZnOlnqgLSN97bV8RNHdpF9wmogdB6xKm5ilICHMnKgRepIBA4HE
J0CeRrFr6LENQbaEU1YGuQ4StithPCSMm3FMNAsvZsM+4ALx+gj7E6NVFgfCyium
oeUyMjqzLE4Ddm3cvelMc0I0hnr7noyzYML6jJaed0XMt5V1Ln/2zUOLfLMc6x9a
SU/mzFnnF2ieqR32M0OijS5E9EHOR46clQlxAQBOdI9qtFTtMYFwe7FIEbvQhJI=
=YI77
-----END PGP SIGNATURE-----
Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging
More migration fixes and more record/replay preparations. Also moves
the sdhci-pci device id to make space for the rocker device.
# gpg: Signature made Sat 03 Jan 2015 08:22:36 GMT using RSA key ID 78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg: aka "Paolo Bonzini <pbonzini@redhat.com>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4 E2F7 7E15 100C CD36 69B1
# Subkey fingerprint: F133 3857 4B66 2389 866C 7682 BFFB D25F 78C7 AE83
* remotes/bonzini/tags/for-upstream:
pci: move REDHAT_SDHCI device ID to make room for Rocker
block/iscsi: fix uninitialized variable
pckbd: set bits 2-3-6-7 of the output port by default
serial: refine serial_thr_ipending_needed
gen-icount: check cflags instead of use_icount global
translate: check cflags instead of use_icount global
cpu-exec: add a new CF_USE_ICOUNT cflag
target-ppc: pass DisasContext to SPR generator functions
atomic: fix position of volatile qualifier
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Pavel Dovgalyuk <pavel.dovgaluk@ispras.ru>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Add instructions of RR1 opcode format, that have 0xb3 as first opcode.
Add helper functions mulh, mulmh and mulrh, that compute multiplication,
with multiprecision (mulmh) or rounding (mulrh) of 4 halfwords, being either low or high parts
of two 32 bit regs.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Fix gen_mtcr using wrong register.
Fix gen_mtcr/mfcr using sign extended offsets.
Fix B format insn using not sign extendend offsets.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Some of the 1.6 ISA instructions were still missing. So let's add them.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add instructions of RR opcode format, that have 0x4b as the first opcode.
Add helper functions:
* parity: Calculates the parity bits for every byte of a 32 int.
* bmerge/bsplit: Merges two regs into one bitwise/Splits one reg into two bitwise.
* unpack: unpack a IEEE 754 single precision floating point number as exponent and mantissa.
* dvinit_b_13/131: (ISA v1.3/v1.31)Prepare operands for a divide operation,
where the quotient result is guaranteed to fit into 8 bit.
* dvinit_h_13/131: (ISA v1.3/v1.31)Prepare operands for a divide operation,
where the quotient result is guaranteed to fit into 16 bit.
OPCM_32_RR_FLOAT -> OPCM_32_RR_DIVIDE.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add instructions of RR opcode format, that have 0x1 as the first opcode.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add instructions of RR opcode format, that have 0xf as the first opcode.
Add helper functions:
* clo/z/s: Counts leading ones/zeros/signs.
* clo/z/s_h: Count leading ones/zeros/signs in two haflwords.
* sh/_h: Shifts one/two word/hwords.
* sha/_h: Shifts one/two word/hwords arithmeticly.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add instructions of RR opcode format, that have 0xb as the first opcode.
Add helper functions, for hword and byte arithmetics:
* add_h_ssov/suov: Add two halfword and saturate on overflow.
* sub_h_ssov/suov: Sub two halfword and saturate on overflow.
* absdif_h_ssov: Compute absolute difference for halfwords and saturate on overflow.
* abs_h_ssov/suov: Compute absolute value for two halfwords and saturate on overflow.
* abs_b/h: Compute absolute value for four/two bytes/halfwords
* absdif_b/h: Compute absolute difference for four/two bytes/halfwords
* add_b/h: Add four/two bytes/halfwords.
* sub_b/h: Sub four/two bytes/halfwords.
* eq_b/h: Compare four/two bytes/halfwords with four/two bytes/halfwords on
equality and set all bits of to either one ore zero.
* eqany_b/h: Compare four/two bytes/halfwords with four/two bytes/halfwords on equality.
* lt_b/bu/h/hu: Compare four/two bytes/halfwords with four/two bytes/halfwords
on less than signed and unsigned.
* max_b/bu/h/hu: Calculate max for four/two bytes/halfwords signed and unsigned.
* min_b/bu/h/hu: Calculate min for four/two bytes/halfwords signed and unsigned.
Add helper function abs_ssov, that computes the absolute value for a 32 bit integer and saturates on overflow.
Add microcode generator functions:
* gen_sub_CC: Caluclates sub and sets the carry bit.
* gen_subc_CC: Caluclates sub and carry and sets the carry bit
* gen_abs: Compute absolute value for a 32 bit integer.
* gen_cond_w: Compares two 32 bit values on cond and sets result either zero or all bits one.
OPC2_32_RR_MIN switched with OPC2_32_RR_MIN_U.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Those makros are exclusively used for 32 bit arithmetics and won't work for
16 bit with two halfwords. So lets get rid of the len parameter and make them
always use 32 bit. Now no token pasting is needed anymore and they can be
regular functions.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
The mask is actually 7 bit long, instead of 6, so the expression checking
for JNZ.T is always false. Let's make the mask 1 bit wider.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Now using psw_read() to retrieve the status bits correctly.
Signed-off-by: Alex Zuepke <alexander.zuepke@hs-rm.de>
Reviewed-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Add instructions of RCR opcode format.
Add helper for madd32/64_ssov and madd32/64_suov.
Add helper for msub32/64_ssov and msub32/64_suov.
Add microcode generator function madd/msub for 32bit and 64bit, which calculate a mul and a add/sub.
OPC2_32_RCR_MSUB_U_32 -> OPC2_32_RCR_MSUB_U_32.
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Add instructions of RLC opcode format.
Add helper psw_write/read.
Add microcode generator gen_mtcr/mfcr, which loads/stores a value to a core special function register, which are defined in csfr.def
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Richard Henderson <rth@twiddle.net>
