cvs magma and hal-linux-2.4.30-i386-r18c1.patch comments
Rus V. Brushkoff
rus at SoyuzKT.Od.UA
Fri Jul 8 07:56:11 CEST 2005
:> 1. The rtai_up module has undefined symbol do_signal, while
:> USE_LINUX_SYSCALL is undefined in the base/include/asm-i386/rtai_lxrt.h
:> (by default in the current cvs).
:With magma there is a must: use just ADEOS patches coming with it. If
:that is done and the missing symbol remain than I've done something
:strange in my commits.
All the comments are related to the adeos patch coming with magma cvs. By
the way it is placed in unusual dir base/ instead of usual
:> 2. In base/include/rtai_nam2num.h lines 58 and 106 the unsigned sizeof
:> value is compared with signed variables which produces compiler warnings.
:I'll look at the above more closely as I've no such warnings and "cvs
:diff -r HEAD" shows no differences between my copy and the official one.
Yes, such warnings someone can see if magma compiled with gcc 3.4.x, I
guess you use gcc-3.3.x, so you do not see these warnings - the type of
sizeof is _always_ unsigned.
:FYI: the i386 part of MAGMA is now under heavy strees testing (and is
:passing it). Daniele has decided to align x86_64 to 2.6.12 and that is
:holding another experimental release for a while.
I've reviewed all released 2.6.x adeos patches, starting nearly from
2.6.5 all them is unstable for me on recent hardware, so for now I
personally revert to 2.4.x ones.
What is interesting - ipipe patches introduced recently by Philippe seems
more stable than adeos ones.
More information about the Rtai