Op 17 okt. 2011, om 18:49 heeft Saul Wold het volgende geschreven:

> On 10/17/2011 08:59 AM, Koen Kooi wrote:
>> 
>> Op 17 okt. 2011, om 17:25 heeft Koen Kooi het volgende geschreven:
>> 
>>> | checking linking as -lgssapi -lkrb5 -ldes -lcrypt -lasn1 -lroken 
>>> -lcom_err... no
>>> | checking linking as -lgssapi -lkrb5 -lcrypto -lcrypt -lasn1 -lroken 
>>> -lcom_err... no
>>> | checking linking as -lgss -lkrb5... no
>>> | configure: error: could not determine proper GSSAPI linkage
>>> | + bbfatal 'oe_runconf failed'
>>> | + echo 'ERROR: oe_runconf failed'
>>> 
>>> Signed-off-by: Koen Kooi<[email protected]>
>>> ---
>>> meta/recipes-connectivity/bind/bind_9.8.1.bb |    5 +++--
>>> 1 files changed, 3 insertions(+), 2 deletions(-)
>>> 
>>> diff --git a/meta/recipes-connectivity/bind/bind_9.8.1.bb 
>>> b/meta/recipes-connectivity/bind/bind_9.8.1.bb
>>> index 330acdb..af473e9 100644
>>> --- a/meta/recipes-connectivity/bind/bind_9.8.1.bb
>>> +++ b/meta/recipes-connectivity/bind/bind_9.8.1.bb
>>> @@ -6,10 +6,10 @@ LICENSE = "ISC&  BSD"
>>> LIC_FILES_CHKSUM = "file://COPYRIGHT;md5=0fbe2a3ab3c68ac3fea3cad13093877c"
>>> 
>>> DEPENDS = "openssl"
>>> -PR = "r0"
>>> +PR = "r1"
>>> 
>>> SRC_URI = "ftp://ftp.isc.org/isc/bind9/${PV}/${BPN}-${PV}.tar.gz \
>>> -      file://conf.patch \
>>> +           file://conf.patch \
>>>            file://cross-build-fix.patch \
>>>        "
>>> 
>>> @@ -19,6 +19,7 @@ SRC_URI[sha256sum] = 
>>> "02285dc429cb2a6687a1b2446e9ee22c1df27f2577225b05be5092395e
>>> # --enable-exportlib is necessary for building dhcp
>>> EXTRA_OECONF = " --enable-ipv6=no --with-randomdev=/dev/random 
>>> --disable-threads \
>>>                  --disable-devpoll --disable-epoll --with-gost=no \
>> 
>> Not related to my patch: if someone can explain we we disable epoll I would 
>> appreciate it, since we generally want it enabled and I think we're using 
>> recent enough kernels (2.6.26+) where epoll finally got hooked up for arm
>> 
> Koen,
> 
> Thanks for pointing this out, a little code spelunking showed that Holger 
> added this back in '08, you might be right about epoll at this point.  Have 
> you tested that we can pass the configure stage and build a working bind?

The thing will epoll is that you can have 2 failure modes:

1) old linux-libc-headers -> build fail
2) recent linux-libc-headers, old kernel -> undefined behaviour

Does anyone have a testcase for epoll in bind?

> On your gssapi, just curious under what OS conditions you found this, I 
> wonder why it passed on multiple systems that I built it on.

Debian testing, ubuntu 10.04 and fedora 15 building any machine with angstrom 
made it fail.

regards,

Koen
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core

Reply via email to