On Monday 04 February 2008, Frans Pop wrote: > There is one issue, which I will detail in a follow-up mail to > debian-boot only. The short summary is that for arm the etchnhalf kernel > meta packages are currently not considered "installable", so as things > stand now the above would not work for arm. The problem is also valid for armeb, armel and m68k, but those arches are not really relevant for Etch+1/2. Right, so this is what happens. Current kernel selection scripts use one of these formats to test whether a kernel is usable (note quoting): 1) expr "$1" : '.*-<flavor>.*' (or with double quoting if <flavor> is a var) 2) expr "$1" : '.*-<flavor>' 3) expr "$1" : '.*-<flavor>$' OR expr "$1" : ".*-<flavor>\$" Tests for a lot of arches look like 1), some like 2). But effectively those are identical, which is confusing. The "problem" arches use 3). With tests 1) and 2), basically _any_ postfix after the flavor is allowed, which could possibly be considered a bug but I do now make use of that fact in the etchnhalf patches. To fix this issue for arm, we could consider the following patch: Index: arm.sh =================================================================== --- arm.sh (revision 51164) +++ arm.sh (working copy) @@ -18,9 +18,11 @@ # Netwinder subarch uses footbridge kernel flavor if [ "$2" = "netwinder" ]; then if expr "$1" : ".*-footbridge\$" >/dev/null; then return 0; fi + if expr "$1" : ".*-footbridge-.*\$" >/dev/null; then return 0; fi fi # Subarchitecture must match exactly if expr "$1" : ".*-$2\$" >/dev/null; then return 0; fi + if expr "$1" : ".*-$2-.*\$" >/dev/null; then return 0; fi return 1 } Comments from arm porters welcome. Is there any reason to tighten the test for other arches in line with that? Cheers, FJP
Attachment:
signature.asc
Description: This is a digitally signed message part.