You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Summary:
In low pointer builds we use the 32 bit Func* as an ID instead of computing separate unique ID numbers for each function. This works fine in repo mode but in non-repo mode it means that FuncId is no longer unique (e.g. if I free a Func and allocate a new one it may be allocated at the same address and therefore share the same FuncId).
This is mostly a theoretical concern as we don't run non-repo low pointer builds anywhere but we do have some tests that depend on being able to calculate unique ProfPrologueIDs in all build configurations. To fix that without expanding the Func structure store a side table of IDs in non-repo mode.
Reviewed By: ricklavoie
Differential Revision: D73924655
fbshipit-source-id: 3495f3ab3c8b941f5b95913e2e7dad21923673e3
0 commit comments