On 11/06/2019 04:43, David Gibson wrote: > On Fri, Jun 07, 2019 at 09:28:49AM -0500, Richard Henderson wrote: >> On 6/7/19 9:09 AM, Laurent Vivier wrote: >>> On 07/06/2019 11:29, Laurent Vivier wrote: >>>> On 29/05/2019 08:49, David Gibson wrote: >>>>> From: Richard Henderson <richard.hender...@linaro.org> >>>>> >>>>> The gvec expanders take care of masking the shift amount >>>>> against the element width. >>>>> >>>>> Signed-off-by: Richard Henderson <richard.hender...@linaro.org> >>>>> Message-Id: <20190518191430.21686-2-richard.hender...@linaro.org> >>>>> Signed-off-by: David Gibson <da...@gibson.dropbear.id.au> >>>>> --- >>>>> target/ppc/helper.h | 12 ---------- >>>>> target/ppc/int_helper.c | 37 ----------------------------- >>>>> target/ppc/translate/vmx-impl.inc.c | 24 +++++++++---------- >>>>> 3 files changed, 12 insertions(+), 61 deletions(-) >>>> >>>> This patch introduces a regressions >>>> with Fedora 29 guest: >>>> >>>> - during kernel boot: >>>> >>>> [ 40.397876] crypto_register_alg 'aes' = 0 >>>> [ 40.577517] crypto_register_alg 'cbc(aes)' = 0 >>>> [ 40.743576] crypto_register_alg 'ctr(aes)' = 0 >>>> [ 41.061379] alg: skcipher: Test 1 failed (invalid result) on encryption >>>> for p8_aes_xts >>>> [ 41.062054] 00000000: 91 7c f6 9e bd 68 b2 ec 9b 9f e9 a3 ea dd a6 92 >>>> [ 41.062163] 00000010: 98 10 35 57 5e dc 36 1e 9a f7 bc ba 39 f2 5c eb >>>> [ 41.062834] crypto_register_alg 'xts(aes)' = 0 >>>> [ 41.077358] alg: hash: Test 2 failed for p8_ghash >>>> [ 41.077553] 00000000: 5f 89 ab f7 20 57 20 57 20 57 20 57 20 57 20 57 >>>> >>>> - with libssl: >>>> >>>> # curl -o /dev/null https://www.google.com >>>> % Total % Received % Xferd Average Speed Time Time Time >>>> Current >>>> Dload Upload Total Spent Left >>>> Speed >>>> 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- >>>> 0 >>>> curl: (35) error:1408F119:SSL routines:ssl3_get_record:decryption failed >>>> or bad record mac >>>> >>>> [before, this one fails with: >>>> curl: (35) error:04091068:rsa routines:int_rsa_verify:bad signature ] >>>> >>>> If I revert this patch on top of 0d74f3b427 + "target/ppc: Fix lxvw4x, >>>> lxvh8x and lxvb16x", all works fine. >>>> >>>> Thanks, >>>> Laurent >>>> >>> >>> This seems to fix the problem: >>> >>> diff --git a/accel/tcg/tcg-runtime-gvec.c b/accel/tcg/tcg-runtime-gvec.c >>> index 3b6052fe97..6f0709b307 100644 >>> --- a/accel/tcg/tcg-runtime-gvec.c >>> +++ b/accel/tcg/tcg-runtime-gvec.c >>> @@ -874,7 +874,7 @@ void HELPER(gvec_sar8v)(void *d, void *a, void *b, >>> uint32_t desc) >>> intptr_t oprsz = simd_oprsz(desc); >>> intptr_t i; >>> >>> - for (i = 0; i < oprsz; i += sizeof(vec8)) { >>> + for (i = 0; i < oprsz; i += sizeof(uint8_t)) { >>> uint8_t sh = *(uint8_t *)(b + i) & 7; >>> *(int8_t *)(d + i) = *(int8_t *)(a + i) >> sh; >>> } >> >> Grr. I really really need to come up with a solution for testing that allows >> me to test paths that the host cpu would not ordinarily take. This bug is >> hidden on a host with AVX2. >> >> Thanks for the digging. > > Can one of you send this fix formally with a S-o-b and so forth?
I'm going to send it. Thanks, Laurent
