From: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
To: Richard Henderson <richard.henderson@linaro.org>, qemu-devel@nongnu.org
Cc: qemu-riscv@nongnu.org, alistair.francis@wdc.com,
bmeng@tinylab.org, liwei1518@gmail.com,
zhiwei_liu@linux.alibaba.com, palmer@rivosinc.com,
philmd@linaro.org
Subject: Re: [PATCH for-9.0 v13 4/8] target/riscv: always clear vstart in whole vec move insns
Date: Wed, 13 Mar 2024 18:39:22 -0300 [thread overview]
Message-ID: <1fc1db23-6942-4968-8978-40999c50936c@ventanamicro.com> (raw)
In-Reply-To: <87936b8f-4900-4a5f-a63d-266c1550c9a1@linaro.org>
On 3/13/24 18:16, Richard Henderson wrote:
> On 3/13/24 09:30, Daniel Henrique Barboza wrote:
>> These insns have 2 paths: we'll either have vstart already cleared if
>> vstart_eq_zero or we'll do a brcond to check if vstart >= maxsz to call
>> the 'vmvr_v' helper. The helper will clear vstart if it executes until
>> the end, or if vstart >= vl.
>>
>> However, if vstart >= maxsz, the helper will be skipped, and vstart
>> won't be cleared since the helper is being responsible from doing it.
>>
>> We want to make the helpers responsible to manage vstart, including
>> these corner cases, precisely to avoid these situations. Move the vstart
>>> = maxsz cond to the helper, and be sure to clear vstart if that
>> happens. This way we're now 100% sure that vstart is being clearer in
>> the end of the execution, regardless of the path taken.
>>
>> Fixes: f714361ed7 ("target/riscv: rvv-1.0: implement vstart CSR")
>> Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
>> ---
>> target/riscv/insn_trans/trans_rvv.c.inc | 3 ---
>> target/riscv/vector_helper.c | 5 +++++
>> 2 files changed, 5 insertions(+), 3 deletions(-)
>>
>> diff --git a/target/riscv/insn_trans/trans_rvv.c.inc b/target/riscv/insn_trans/trans_rvv.c.inc
>> index 8c16a9f5b3..52c26a7834 100644
>> --- a/target/riscv/insn_trans/trans_rvv.c.inc
>> +++ b/target/riscv/insn_trans/trans_rvv.c.inc
>> @@ -3664,12 +3664,9 @@ static bool trans_##NAME(DisasContext *s, arg_##NAME * a) \
>> vreg_ofs(s, a->rs2), maxsz, maxsz); \
>> mark_vs_dirty(s); \
>> } else { \
>> - TCGLabel *over = gen_new_label(); \
>> - tcg_gen_brcondi_tl(TCG_COND_GEU, cpu_vstart, maxsz, over); \
>> tcg_gen_gvec_2_ptr(vreg_ofs(s, a->rd), vreg_ofs(s, a->rs2), \
>> tcg_env, maxsz, maxsz, 0, gen_helper_vmvr_v); \
>> mark_vs_dirty(s); \
>> - gen_set_label(over); \
>> } \
>> return true; \
>> } \
>> diff --git a/target/riscv/vector_helper.c b/target/riscv/vector_helper.c
>> index b4360dbd52..7260a5972b 100644
>> --- a/target/riscv/vector_helper.c
>> +++ b/target/riscv/vector_helper.c
>> @@ -5163,6 +5163,11 @@ void HELPER(vmvr_v)(void *vd, void *vs2, CPURISCVState *env, uint32_t desc)
>> VSTART_CHECK_EARLY_EXIT(env);
>> + if (env->vstart >= maxsz) {
>> + env->vstart = 0;
>> + return;
>> + }
>
> Did you need the VSTART_CHECK_EARLY_EXIT here then?
> It certainly seems like the vstart >= vl check is redundant...
Hmm right. I thought about maxsz being calculated via vlmax, and vlmax not necessarily
being == vl ... but vlmax will be always >= vl. So the vstart >= vl check s supersed
by vstart >= maxsz.
I'll re-send. Thanks,
Daniel
>
>
> r~
next prev parent reply other threads:[~2024-03-13 21:39 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-03-13 19:30 [PATCH for-9.0 v13 0/8] riscv: set vstart_eq_zero on vector insns Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 1/8] target/riscv/vector_helper.c: set vstart = 0 in GEN_VEXT_VSLIDEUP_VX() Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 2/8] trans_rvv.c.inc: set vstart = 0 in int scalar move insns Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 3/8] target/riscv/vector_helpers: do early exit when vstart >= vl Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 4/8] target/riscv: always clear vstart in whole vec move insns Daniel Henrique Barboza
2024-03-13 21:16 ` Richard Henderson
2024-03-13 21:39 ` Daniel Henrique Barboza [this message]
2024-03-13 19:30 ` [PATCH for-9.0 v13 5/8] target/riscv: remove 'over' brconds from vector trans Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 6/8] trans_rvv.c.inc: remove redundant mark_vs_dirty() calls Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 7/8] target/riscv: enable 'vstart_eq_zero' in the end of insns Daniel Henrique Barboza
2024-03-13 19:30 ` [PATCH for-9.0 v13 8/8] target/riscv/vector_helper.c: optimize loops in ldst helpers Daniel Henrique Barboza
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1fc1db23-6942-4968-8978-40999c50936c@ventanamicro.com \
--to=dbarboza@ventanamicro.com \
--cc=alistair.francis@wdc.com \
--cc=bmeng@tinylab.org \
--cc=liwei1518@gmail.com \
--cc=palmer@rivosinc.com \
--cc=philmd@linaro.org \
--cc=qemu-devel@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=zhiwei_liu@linux.alibaba.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).