RFR: 8324833: Signed integer overflows in ABS [v2]

[Dean Long]

On Wed, 31 Jan 2024 15:55:25 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

>> See the details in the bug. I think current `ABS` implementation is beyond repair, and we should just switch to `uabs`.
>> 
>> Additional testing:
>>  - [x] Linux x86_64 fastdebug, `all` with `-ftrapv` (now fully passes!)
>>  - [x] Linux x86_64 fastdebug, `all`
>
> Aleksey Shipilev has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Unnecessary comment

src/hotspot/share/jvmci/jvmciCompilerToVM.cpp line 1063:

> 1061:     int64_t off_low = (int64_t)target_addr - ((int64_t)CodeCache::low_bound() + sizeof(int));
> 1062:     int64_t off_high = (int64_t)target_addr - ((int64_t)CodeCache::high_bound() + sizeof(int));
> 1063:     return MAX2(ABS(off_low), ABS(off_high));

Wouldn't this trigger a warning with -Wsign-conversion enabled?

src/hotspot/share/opto/ifnode.cpp line 1978:

> 1976:           // "x - y" -> must add one to the difference for number of elements in [x,y]
> 1977:           const jlong diff = (jlong)MIN2(offset2, off_lo) - (jlong)MAX2(offset2, off_hi);
> 1978:           if (uabs(diff) < maximum_number_of_min_max_interval_indices) {

unsigned vs signed comparison.  Maybe the rhs should be unsigned as well?

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/17617#discussion_r1475100564
PR Review Comment: https://git.openjdk.org/jdk/pull/17617#discussion_r1475102253