[v2] target/arm: assorted mte fixes

[PATCH v2 0/6] target/arm: assorted mte fixes

Posted by Richard Henderson 9 months, 3 weeks ago

The first patch is unchanged from

Supercedes: <20240131003557.176486-1-richard.henderson@linaro.org>

while the remaining patches replace

Supercedes: <20240205023948.25476-1-richard.henderson@linaro.org>

While digging through Gustavo's test case, wondering why it
should be failing at all, I finally noticed that we weren't
overflowing MTEDESC.SIZEM1, but underflowing (-1).  Oops.

But I did find a few other points by inspection where we
weren't properly handling or supplying MTEDESC.


r~


Richard Henderson (6):
  linux-user/aarch64: Extend PR_SET_TAGGED_ADDR_CTRL for FEAT_MTE3
  target/arm: Fix nregs computation in do_ld_zpa
  target/arm: Adjust and validate mtedesc sizem1
  target/arm: Split out make_svemte_desc
  target/arm: Handle mte in do_ldrq, do_ldro
  target/arm: Fix SVE/SME gross MTE suppression checks

 linux-user/aarch64/target_prctl.h | 25 +++++-----
 target/arm/internals.h            |  2 +-
 target/arm/tcg/translate-a64.h    |  2 +
 target/arm/tcg/sme_helper.c       |  8 ++--
 target/arm/tcg/sve_helper.c       | 12 ++---
 target/arm/tcg/translate-sme.c    | 15 ++----
 target/arm/tcg/translate-sve.c    | 80 ++++++++++++++++++-------------
 7 files changed, 78 insertions(+), 66 deletions(-)

-- 
2.34.1

Re: [PATCH v2 0/6] target/arm: assorted mte fixes

Posted by Gustavo Romero 9 months, 3 weeks ago

Hi Richard,

On 2/6/24 12:05 AM, Richard Henderson wrote:
> The first patch is unchanged from
> 
> Supercedes: <20240131003557.176486-1-richard.henderson@linaro.org>
> 
> while the remaining patches replace
> 
> Supercedes: <20240205023948.25476-1-richard.henderson@linaro.org>
> 
> While digging through Gustavo's test case, wondering why it
> should be failing at all, I finally noticed that we weren't
> overflowing MTEDESC.SIZEM1, but underflowing (-1).  Oops.
> 
> But I did find a few other points by inspection where we
> weren't properly handling or supplying MTEDESC.
> 
> 
> r~
> 
> 
> Richard Henderson (6):
>    linux-user/aarch64: Extend PR_SET_TAGGED_ADDR_CTRL for FEAT_MTE3
>    target/arm: Fix nregs computation in do_ld_zpa
>    target/arm: Adjust and validate mtedesc sizem1
>    target/arm: Split out make_svemte_desc
>    target/arm: Handle mte in do_ldrq, do_ldro
>    target/arm: Fix SVE/SME gross MTE suppression checks
> 
>   linux-user/aarch64/target_prctl.h | 25 +++++-----
>   target/arm/internals.h            |  2 +-
>   target/arm/tcg/translate-a64.h    |  2 +
>   target/arm/tcg/sme_helper.c       |  8 ++--
>   target/arm/tcg/sve_helper.c       | 12 ++---
>   target/arm/tcg/translate-sme.c    | 15 ++----
>   target/arm/tcg/translate-sve.c    | 80 ++++++++++++++++++-------------
>   7 files changed, 78 insertions(+), 66 deletions(-)
> 

Tested-by: Gustavo Romero <gustavo.romero@linaro.org>


Thanks!

Re: [PATCH v2 0/6] target/arm: assorted mte fixes

Posted by Peter Maydell 9 months, 3 weeks ago

On Tue, 6 Feb 2024 at 03:07, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> The first patch is unchanged from
>
> Supercedes: <20240131003557.176486-1-richard.henderson@linaro.org>
>
> while the remaining patches replace
>
> Supercedes: <20240205023948.25476-1-richard.henderson@linaro.org>
>
> While digging through Gustavo's test case, wondering why it
> should be failing at all, I finally noticed that we weren't
> overflowing MTEDESC.SIZEM1, but underflowing (-1).  Oops.
>
> But I did find a few other points by inspection where we
> weren't properly handling or supplying MTEDESC.
>

Should some or all of this patchset be cc: qemu-stable ?

thanks
-- PMM

Re: [PATCH v2 0/6] target/arm: assorted mte fixes

Posted by Richard Henderson 9 months, 3 weeks ago

On 2/7/24 00:54, Peter Maydell wrote:
> On Tue, 6 Feb 2024 at 03:07, Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> The first patch is unchanged from
>>
>> Supercedes: <20240131003557.176486-1-richard.henderson@linaro.org>
>>
>> while the remaining patches replace
>>
>> Supercedes: <20240205023948.25476-1-richard.henderson@linaro.org>
>>
>> While digging through Gustavo's test case, wondering why it
>> should be failing at all, I finally noticed that we weren't
>> overflowing MTEDESC.SIZEM1, but underflowing (-1).  Oops.
>>
>> But I did find a few other points by inspection where we
>> weren't properly handling or supplying MTEDESC.
>>
> 
> Should some or all of this patchset be cc: qemu-stable ?

All of it, I expect.


r~