Skip to contentSkip to conversation

In consistancy in the defition of BBFILE_COLLECTIONS in layer.conf

DISCUSSION Jeremy Puhlman 4 participants 5 messages May 26, 2011

lore
Share & tools
lore
#0 lore raw

Opps sent from wrong account.

I was working with some layerinfo code, and I noticed that the values of BBFILE_COLLECTIONS in layer.conf in the meta-intel repo are kind of all over the map:

for each in */conf/layer.conf; do cat $each; done | grep BBFILE_COLLECTIONS
BBFILE_COLLECTIONS_crownbay += "crownbay" BBFILE_COLLECTIONS_crownbay-noemgd += "crownbay-noemgd" BBFILE_COLLECTIONS_fishriver += "fishriver" BBFILE_COLLECTIONS_jasperforest += "jasperforest" BBFILE_COLLECTIONS_sugarbay += "sugarbay"

For the above layers when I just list out oe-core and one of the above, the results from BBFILE_COLLECTIONS ends up
just being what ever was set above.

BBFILE_COLLECTIONS += "n450"
BBFILE_COLLECTIONS += "emenlow"

When using these layers, the final BBFILE_COLLECTIONS is the list of all the layers as expected.

Was just curious if there was a intentional reason why the difference(i.e. is there some code that I am missing that fixes this up)? -- Jeremy Puhlman Montavista Sofware, LLC.

#1 lore raw
Quoted 16 lines · Jeremy Puhlman

Opps sent from wrong account.

I was working with some layerinfo code, and I noticed that the values of BBFILE_COLLECTIONS in layer.conf in the meta-intel repo are kind of all over the map:

for each in */conf/layer.conf; do cat $each; done | grep BBFILE_COLLECTIONS
BBFILE_COLLECTIONS_crownbay += "crownbay" BBFILE_COLLECTIONS_crownbay-noemgd += "crownbay-noemgd" BBFILE_COLLECTIONS_fishriver += "fishriver" BBFILE_COLLECTIONS_jasperforest += "jasperforest" BBFILE_COLLECTIONS_sugarbay += "sugarbay"

For the above layers when I just list out oe-core and one of the above, the results from BBFILE_COLLECTIONS ends up
just being what ever was set above.

Quoted 8 lines

BBFILE_COLLECTIONS += "n450"
BBFILE_COLLECTIONS += "emenlow"

When using these layers, the final BBFILE_COLLECTIONS is the list of all the layers as expected.

Was just curious if there was a intentional reason why the difference(i.e. is there some code that I am missing that fixes this up)?

No, not intentional - looks like a case of an overzealous overriding typo that propagated. I'll cook up a patch to make them all just 'BBFILE_COLLECTIONS += xxxx'

Tom

Quoted 6 lines

-- Jeremy Puhlman Montavista Sofware, LLC.

_______________________________________________
yocto mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/yocto

#2 lore raw
Quoted 7 lines

No, not intentional - looks like a case of an overzealous overriding typo that propagated. I'll cook up a patch to make them all just 'BBFILE_COLLECTIONS += xxxx'

Tom

Good deal. Just wanted to double check. Thanks for the quick response.

Signature · 2 lines
Jeremy Puhlman
Montavista Sofware, LLC.
#3 lore raw
In reply to Tom Zanussi
Quoted 23 lines · Tom Zanussi
Quoted 22 lines · Jeremy Puhlman

Opps sent from wrong account.

I was working with some layerinfo code, and I noticed that the values of BBFILE_COLLECTIONS in layer.conf in the meta-intel repo are kind of all over the map:

for each in */conf/layer.conf; do cat $each; done | grep BBFILE_COLLECTIONS
BBFILE_COLLECTIONS_crownbay += "crownbay" BBFILE_COLLECTIONS_crownbay-noemgd += "crownbay-noemgd" BBFILE_COLLECTIONS_fishriver += "fishriver" BBFILE_COLLECTIONS_jasperforest += "jasperforest" BBFILE_COLLECTIONS_sugarbay += "sugarbay"

For the above layers when I just list out oe-core and one of the above, the results from BBFILE_COLLECTIONS ends up
just being what ever was set above.

BBFILE_COLLECTIONS += "n450"
BBFILE_COLLECTIONS += "emenlow"

When using these layers, the final BBFILE_COLLECTIONS is the list of all the layers as expected.

Quoted 6 lines
Quoted 2 lines

Was just curious if there was a intentional reason why the difference(i.e. is there some code that I am missing that fixes this up)?

No, not intentional - looks like a case of an overzealous overriding typo that propagated. I'll cook up a patch to make them all just 'BBFILE_COLLECTIONS += xxxx'

Using machine override has one small benefit

when I have meta-nokia layer which has only recipes with
COMPATIBLE_MACHINE = "nokia900" and I'm building different machine (like
om-gta02) I get this warning:
WARNING: No bb files matched BBFILE_PATTERN_nokia-layer '^/OE/shr-core/meta-smartphone/meta-nokia/' because all bb files from this layer are ignored because of
COMPATIBLE_MACHINE

if I change it like this
-BBFILE_COLLECTIONS += "nokia-layer" +BBFILE_COLLECTIONS_nokia900 += "nokia-layer"
maybe with more overrides per layer.conf if it supports multiple machines, the warning is gone, but seems like sstate checksums are different again, because during test I changed only this and another gcc rebuild is in runqueue now :/.

Signature · 1 line
Martin 'JaMa' Jansa     jabber: [email protected]
#4 lore raw
Quoted 8 lines · Martin Jansa

Using machine override has one small benefit

when I have meta-nokia layer which has only recipes with
COMPATIBLE_MACHINE = "nokia900" and I'm building different machine (like
om-gta02) I get this warning:
WARNING: No bb files matched BBFILE_PATTERN_nokia-layer '^/OE/shr-core/meta-smartphone/meta-nokia/' because all bb files from this layer are ignored because of
COMPATIBLE_MACHINE

I think there is a change on bitbake master which should fix some of the cases where this warning was appearing (where a layer only contained .bbappend files).

Quoted 7 lines

if I change it like this

(quoted diff) +1 -1
-BBFILE_COLLECTIONS += "nokia-layer"
+BBFILE_COLLECTIONS_nokia900 += "nokia-layer"

maybe with more overrides per layer.conf if it supports multiple machines, the warning is gone, but seems like sstate checksums are different again, because during test I changed only this and another gcc rebuild is in runqueue now :/.

Ouch. If you can figure out what changed we can try and fix that. FWIW, I also merged the signature generation data writing at stamp creation time (with fixes).

Cheers,

Richard