Alex Coplan <alex.cop...@arm.com> writes: > Hi Richard, > > On 21/04/2021 13:05, Richard Sandiford wrote: >> Alex Coplan <alex.cop...@arm.com> writes: >> > Hi Richard, >> > >> > On 15/04/2021 18:45, Richard Sandiford wrote: >> >> Looks good in general, but like you say, it's GCC 12 material. >> > >> > Thanks for the review. The attached patch addresses these comments and >> > bootstraps/regtests OK on aarch64-linux-gnu. OK for trunk? >> >> OK, thanks. > > The patch applies cleanly and bootstraps/regtests OK on the GCC 11 > branch. OK to backport there and to 10 and 9 if the same is true for > those branches?
To summarise what we discussed off-list: this initially looked like it was “just” a new DCE optimisation, which is why it seemed like GCC 12 material. However, as Alex pointed out, the unpatched BTI support can generate code whose size is quadratic in the number of switch cases, which is a serious regression whichever way you cut it. So this is OK for backports, and would have been OK during GCC 11 stage 4 too. Thanks, Richard
