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
my earliest injection ( issue #234 ) today restores its hook immediately. there
can be super-early threads when injected from outside (including from
system itself during shutdown and other times) (and really there's nothing
special about the main thread: whoever gets to init routine first is the
initializer and will then be the DR intializer). the idea is to leave the
hook but change it to spin (or if manage to get syscall support, wait) so
we can queue up these threads until DR is fully initialized.
note that in the past we discussed ways to try and avoid any other thread
being schedulable until we were set up, via SetProcessAffinityMask(0) or
sthg, but none of those panned out
this would block windbg's thread as well and if windbg can handle that
would enable debugging at pre-Ldr-init time
From [email protected] on November 29, 2011 10:11:01
my earliest injection ( issue #234 ) today restores its hook immediately. there
can be super-early threads when injected from outside (including from
system itself during shutdown and other times) (and really there's nothing
special about the main thread: whoever gets to init routine first is the
initializer and will then be the DR intializer). the idea is to leave the
hook but change it to spin (or if manage to get syscall support, wait) so
we can queue up these threads until DR is fully initialized.
note that in the past we discussed ways to try and avoid any other thread
being schedulable until we were set up, via SetProcessAffinityMask(0) or
sthg, but none of those panned out
this would block windbg's thread as well and if windbg can handle that
would enable debugging at pre-Ldr-init time
Original issue: http://code.google.com/p/dynamorio/issues/detail?id=627
The text was updated successfully, but these errors were encountered: