Message ID | 20220905200652.d69204581d15c64647da5cd2@otheo.eu |
---|---|
Headers | show |
Series | elf: Fix hwcaps string size overestimation | expand |
* Javier Pello: > Here is a series of patches related to the construction of the hwcaps > power set strings in function _dl_important_hwcaps in elf/dl-hwcaps.c. We deprecated this mechanism in glibc 2.33. I think it's time to remove it in glibc 2.37. Since you are evidently familiar with the code now, would like to implement the removal? Thanks, Florian
> We deprecated this mechanism in glibc 2.33. I think it's time > to remove it in glibc 2.37. > > Since you are evidently familiar with the code now, That may be a bit of a stretch. :-) > would like to implement the removal? What would this removal be? Do you want _dl_important_hwcaps to return only the glibc-hwcaps subdirectories, and not the legacy hwcaps power set subdirectories? That should be doable. In any case, even if hwcaps power set support is removed, I think that at least my first patch sould be applied first. If the removal were to be reverted later, or if somebody looked at the code as it was before the removal for whatever reason, I think that it would be better if it had been left in the best possible shape, at least with the overallocation bug fixed. Regards, Javier
* Javier Pello: >> We deprecated this mechanism in glibc 2.33. I think it's time >> to remove it in glibc 2.37. >> >> Since you are evidently familiar with the code now, > > That may be a bit of a stretch. :-) > >> would like to implement the removal? > > What would this removal be? Do you want _dl_important_hwcaps to > return only the glibc-hwcaps subdirectories, and not the legacy > hwcaps power set subdirectories? That should be doable. Exactly. And in separate commits, update ldconfig and the cache reading logic in elf/dl-cache.c. > In any case, even if hwcaps power set support is removed, I think > that at least my first patch sould be applied first. If the removal > were to be reverted later, or if somebody looked at the code as it > was before the removal for whatever reason, I think that it would > be better if it had been left in the best possible shape, at least > with the overallocation bug fixed. Agreed. Thanks, Florian