Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2004-03-12 09:10:47 +0100 \ Weyer, Wolfram: > Hi, > I am currently integrating drbd_0.6.12 with the MVLCGE3.1, which is based on > the 2.4.20 kernel. > It seems that the SIGHAND_HACK has to be set because of the SIGMASK_LOCK and > RECALC_SIGPENDING defines in drbd_int.h. The definitions for > has_pending_signals and recalc_sigpending_tsk however are already present in > sched.h. I solved this for me by just renaming the define in drbd_main.c to > SIGHAND_HACK2 (and _not_ defining this in drbd_config.h). > I know you cannot take care for all different kernel flavours but may there > are similar constellations arround so that a generic solution makes sense. Thank you. Yes, static inline something should go into .h files. But e.g. Redhat and RH based systems badly chose to have this (at least with their kernel 2.4.22*) in sched.c (not .h), and not exported. The "generic solution" is to get it included in the mainstream kernel, then its the distributors problem, not ours. Well, actually it is never our problem, since we basically only support mainstream kernels. Ok, and sponsor kernels :) But notone runs kernel.org kernels, so we obviously had to support the RH variants, too. Noone ever complained about the Montavista kernel yet... Lars Ellenberg