Re: [PULL v2 05/13] target/rx: CPU definitions

["Philippe Mathieu-Daudé" <philmd@redhat.com>]

On 3/20/20 5:37 PM, Peter Maydell wrote:
> On Fri, 20 Mar 2020 at 16:32, Philippe Mathieu-Daudé <philmd@redhat.com> wrote:
>> -fwrapv is here indeed.
>>
>> I use
>> --extra-cflags=-fsanitize=address,alignment,array-bounds,bool,builtin,enum,float-cast-overflow,float-divide-by-zero,function,integer-divide-by-zero,nonnull-attribute,null,pointer-overflow,return,returns-nonnull-attribute,shift-base,shift-exponent,signed-integer-overflow,unreachable,vla-bound,vptr
> 
> There was a bug in older clang versions where the shift-base
> sanitizer didn't honour -fwrapv:
> https://bugs.llvm.org/show_bug.cgi?id=25552
> 
> https://wiki.qemu.org/Testing#clang_UBSan
> says you can work around the clang bug with -fno-sanitize=shift-base.
> 
> The bug was fixed upstream back in 2016, though, so the
> fix ought to be in clang 4, I think. Are you using an
> old clang version, or has it regressed in newer clang?

I am sorry I am very confused here.

$ clang -v
clang version 8.0.0 (Fedora 8.0.0-3.fc30)

I use scan-build which set CC=ccc-analyzer.

$ ccc-analyzer -v
gcc version 9.2.1 20190827 (Red Hat 9.2.1-1) (GCC)

I use the output of 'make V=1' and run directly ccc-analyzer, I get:

gcc: error: unrecognized argument to ‘-fsanitize=’ option: ‘array-bounds’
gcc: error: unrecognized argument to ‘-fsanitize=’ option: ‘function’

So Clang is used.

I notice ccc-analyzer appends -Wno-shift-negative-value, however if I 
run the 'make V=1' output using clang instead (with all the -fsanitize 
options, -fwrapv, -Wno-shift-negative-value) then no warning are emitted.

So I think this is simply a problem with scan-build/ccc-analyzer, or I 
should tune it more for QEMU.