Re: Re: [Patch][binutils][0/10] Support for Armv8.6-A

[Mihail Ionescu <mihail dot ionescu at foss dot arm dot com>]

Hi Nick,

I've fixed the warning/error and the regressions.
I've attached a zip of all of the diffs to this email and replied with 
the updated ChangeLogs/diffs to the individual patches, although, it
looks like, for some reason, the In-Reply-To initial patch did not work 
and they arrived as new entries instead of replies; sorry for the 
inconvenience.


Regards,
Mihail


On 10/31/2019 12:34 PM, Nick Clifton wrote:
> Hi Mihail,
>
> > This is a patch series to implement support for the Armv8.6-A Architecture[1].
> > Armv8.6-A introduces two new extensions Matix Multiply and BFloat16[2].
>
> With this patch set applied I see a new build time error/warning:
>
>    gas/config/tc-arm.c: In function 'parse_neon_type':
>    gas/config/tc-arm.c:1526:4: error: this statement may fall through [-Werror=implicit-fallthrough=]
>        switch (TOLOWER (*(++ptr)))
>        ^~~~~~
>    gas/config/tc-arm.c:1545:2: note: here
>      default:
>      ^~~~~~~
>
> With this fixed, I then see some new testsuite failures for various
> different configurations of the toolchain:
>
> Checking Binutils in: arm-nacl ... GAS: 1  done
>    GAS REGRESSION: Bfloat 16 extension Thumb
>
> Checking Binutils in: arm-symbianelf ... GAS: 1  LD: 70  BIN: 1  done
>    GAS REGRESSION: Bfloat 16 extension Thumb
>
> Checking Binutils in: aarch64_be-elf ... GAS: 1  done
>    GAS REGRESSION: gas/aarch64/bfloat16
>
> Checking Binutils in: arm-wince-pe ... GAS: 3  done
>    GAS REGRESSION: Big endian bfloat16 literal directives
>    GAS REGRESSION: Little endian bfloat16 literal directives
>    GAS REGRESSION: Bfloat 16 extension Thumb
>
> Would you mind investigating these please ?
>
> Cheers
>    Nick

Attachment: all-patches.tar.gz
Description: application/gzip