[OE-core] a few questions about "COMPATIBLE_MACHINE" variable

[Christopher Larson]

On Mon, Dec 19, 2016 at 8:17 AM, Robert P. J. Day <rpjday at crashcourse.ca>
wrote:

>   as a starting point, COMPATIBLE_MACHINE is processed on a per-recipe
> basis, and if it has no value, then there is no machine restriction
> being applied to that recipe, correct? that's based on this snippet
> from base.bbclass:
>
>   need_machine = d.getVar('COMPATIBLE_MACHINE')
>   if need_machine:
>       import re
>       compat_machines = (d.getVar('MACHINEOVERRIDES') or "").split(":")
>       for m in compat_machines:
>           if re.match(need_machine, m):
>               break
>       else:
>           raise bb.parse.SkipPackage("incompatible with machine %s (not
> in COMPATIBLE_MACHINE)" % d.getVar('MACHINE'))
>
> so far, so good?
>
>   next, the documentation describes the value of that variable as a
> regular expression, so the values are processed as RE patterns, but
> some of the actual uses are confusing. from
> poky/meta/recipes-kernel/linux:
>
> $ grep -r COMPATIBLE_MACHINE *
> linux-dummy.bb:#COMPATIBLE_MACHINE = "your_machine"
> linux-yocto_4.1.bb:COMPATIBLE_MACHINE = "qemuarm|qemuarm64|qemux86|
> qemuppc|qemumips|qemumips64|qemux86-64"
> linux-yocto_4.4.bb:COMPATIBLE_MACHINE = "qemuarm|qemuarm64|qemux86|
> qemuppc|qemumips|qemumips64|qemux86-64"
> linux-yocto_4.8.bb:COMPATIBLE_MACHINE = "qemuarm|qemuarm64|qemux86|
> qemuppc|qemumips|qemumips64|qemux86-64"
> linux-yocto-dev.bb:COMPATIBLE_MACHINE = "(qemuarm|qemux86|qemuppc|
> qemumips|qemumips64|qemux86-64)"
> linux-yocto-rt_4.1.bb:COMPATIBLE_MACHINE = "(qemux86|qemux86-64|qemuarm|
> qemuppc|qemumips)"
> linux-yocto-rt_4.4.bb:COMPATIBLE_MACHINE = "(qemux86|qemux86-64|qemuarm|
> qemuppc|qemumips)"
> linux-yocto-rt_4.8.bb:COMPATIBLE_MACHINE = "(qemux86|qemux86-64|qemuarm|
> qemuppc|qemumips)"
> linux-yocto-tiny_4.1.bb:COMPATIBLE_MACHINE = "(qemux86$)"
> linux-yocto-tiny_4.4.bb:COMPATIBLE_MACHINE = "(qemux86$)"
> linux-yocto-tiny_4.8.bb:COMPATIBLE_MACHINE = "(qemux86$)"
> $
>
>   first, what is best practice -- to use parentheses or not? i'm
> assuming it makes no difference, but it does seem inconsistent and
> could cause some confusion.
>

They’re not needed in a case like this.

  next, if the possibilities in a list are REs, what is the point of
> explicitly listing, say, "qemuarm|qemuarm64"? would not the RE
> "qemuarm" subsume the more explicit "qemuarm64"? same for the other
> architectures. (one could suggest that that entire line could be
> shortened to "... = (^qemu)".)
>

Just qemu would potentially match future qemu machines which aren’t
actually supported, so I don’t think that would be appropriate. It’d match
too much.

  the above seems pretty clear since the following lines would appear
> to say that *only* the qemux86 is compatible:
>
>   linux-yocto-tiny_4.1.bb:COMPATIBLE_MACHINE = "(qemux86$)"
>   linux-yocto-tiny_4.4.bb:COMPATIBLE_MACHINE = "(qemux86$)"
>   linux-yocto-tiny_4.8.bb:COMPATIBLE_MACHINE = "(qemux86$)"
>
> which suggests the following passage from the YP kernel dev manual is
> a bit misleading:
>
>   "You must change it to match a list of the machines that your new
>   recipe supports. For example, to support the qemux86 and qemux86-64
>   machines, use the following form:
>
>        COMPATIBLE_MACHINE = "qemux86|qemux86-64"
>
> and if all this is true, then if you're introducing a new machine, to
> be magnificently pedantic, one should not use:
>
>   COMPATIBLE_MACHINE_machinename = "machinename"
>
> but
>
>   COMPATIBLE_MACHINE_machinename = "^machinename$"
>
> just to play it safe. am i reading all this correctly?
>

Yes, that’s correct, though we need the trailing $ but not the leading ^,
as it’s using re.match, not re.search — meaning it only matches at the
beginning of the string, it doesn’t search the string to find a match.
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openembedded.org/pipermail/openembedded-core/attachments/20161219/85dd2629/attachment-0002.html>