mailarchive of the ptxdist mailing list
 help / color / mirror / Atom feed
From: "Andreas Bießmann" <biessmann@corscience.de>
To: ptxdist@pengutronix.de
Subject: Re: [ptxdist] awk bug in some make dependency scripts?
Date: Tue, 13 Nov 2012 15:35:40 +0100	[thread overview]
Message-ID: <50A25ABC.3050702@corscience.de> (raw)
In-Reply-To: <20121112114733.GS17967@pengutronix.de>

Hi,

On 12.11.2012 12:47, Michael Olbrich wrote:
> Hi,
> 
> On Mon, Nov 12, 2012 at 10:02:06AM +0100, Andreas Bießmann wrote:
>> On 09.11.2012 17:28, Michael Olbrich wrote:
>>> On Fri, Nov 09, 2012 at 03:20:55PM +0100, Andreas Bießmann wrote:
>>>> I've encountered a possible bug in some of the awk scripts generating
>>>> the dependencies. Have not looked deeper into but have this information
>>>> so far.
>>>>
>>>> I use ptxdist-2012.07.0, if I do a clean build on a Debian stable (gawk
>>>> 3.1.7.dfsg-5) the kernel-modules package lack our own modules which
>>>> should be placed in /lib/modules/$(uname -r)/extra.
>>>> If I build the same stuff with the same ptxdist on the same machine but
>>>> use the gawk 4.0.1+dfsg-2~bpo60+1 from debian backports I get the
>>>> modules installed successfully.
>>>>
>>>> Did anyone else see this error before? Some pointers where to start
>>>> digging into it?
>>>
>>> Is the file platform-<myplat>/state/ptx_dgen_deps.post different?
>>
>> yes it is:
>>
>> ---8<---
>> diff -Nrupa /tmp/ptx_dgen_deps.post.old-awk
>> /tmp/ptx_dgen_deps.post.new-awk  | diffstat
>>  ptx_dgen_deps.post.new-awk | 5198
>> ++++++++++++++++++++++-----------------------
>>  1 file changed, 2599 insertions(+), 2599 deletions(-)
>> --->8---
>>
>> On first sight it seems just reordered but will investigate that further.
> 
> Try sorting the files. Most stuff is one-line only anyways and the rest is
> some "ifdef" stuff that should still be readable.

double checked it now. The content of both ptx_dgen_deps.post is the
same after reordering.
I have another strange investigation: I built a smaller testcase to have
faster test results but unfortunately the smaller testcase did always work!
The small Testcase consists only of a kernel plus a simple test module
(and the required host tools). The difference is something like this:

---8<---
% ls -l platform-responder2/state/*.install | wc -l
95
% ls -l platform-test/state/*.install | wc -l
27
--->8---

These following lines show my test-case with the real repository. The
repository is first built completely (so just the rebuild is required).
The rebuild of kernel and datasource-driver is triggered explicitly,
dependency for printer-driver is met implicitly. To show the content of
kernel-modules.ipkg I delete the 'extra' dir in NFS-root before and
check it after the build.
The difference in order of targets built is sticking out and until now
always reproducible (just switch the awk version and do the sequence).

---8<---
with gawk 4.0 (sudo aptitude install gawk/squeeze-backports)
rm -rf platform-responder2/root/lib/modules/3.4.18/extra/
./p drop kernel.compile
./p drop datasource-driver.compile
./p go

target: kernel.compile
target: kernel.install
target: kernel.install.pack
target: kernel.install.post
target: kernel.targetinstall
target: printer-driver.extract.post
target: printer-driver.prepare
target: printer-driver.compile
target: printer-driver.install
target: printer-driver.install.pack
target: printer-driver.install.post
target: printer-driver.targetinstall
target: printer-driver.targetinstall.post
target: datasource-driver.extract.post
target: datasource-driver.prepare
target: datasource-driver.compile
target: datasource-driver.install
target: datasource-driver.install.pack
target: datasource-driver.install.post
target: datasource-driver.targetinstall
target: datasource-driver.targetinstall.post
target: kernel.targetinstall.post

ls -la platform-responder2/root/lib/modules/3.4.18/extra -> OK
----
with gawk 3.1.x (sudo aptitude install gawk/stable)

rm -rf platform-responder2/root/lib/modules/3.4.18/extra/
./p drop kernel.compile
./p drop datasource-driver.compile
./p go

target: kernel.compile
target: kernel.install
target: kernel.install.pack
target: kernel.install.post
target: kernel.targetinstall
target: kernel.targetinstall.post
target: datasource-driver.extract.post
target: datasource-driver.prepare
target: datasource-driver.compile
target: datasource-driver.install
target: datasource-driver.install.pack
target: datasource-driver.install.post
target: datasource-driver.targetinstall
target: datasource-driver.targetinstall.post
target: printer-driver.extract.post
target: printer-driver.prepare
target: printer-driver.compile
target: printer-driver.install
target: printer-driver.install.pack
target: printer-driver.install.post
target: printer-driver.targetinstall
target: printer-driver.targetinstall.post

ls -la platform-responder2/root/lib/modules/3.4.18/extra -> NOT OK
--->8---

Any other pointers?

Best regards

Andreas Bießmann

-- 
ptxdist mailing list
ptxdist@pengutronix.de

  reply	other threads:[~2012-11-13 14:35 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-11-09 14:20 Andreas Bießmann
2012-11-09 16:28 ` Michael Olbrich
2012-11-12  9:02   ` Andreas Bießmann
2012-11-12 11:47     ` Michael Olbrich
2012-11-13 14:35       ` Andreas Bießmann [this message]
2012-11-13 20:02         ` Michael Olbrich
2012-11-14  9:08           ` Andreas Bießmann

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=50A25ABC.3050702@corscience.de \
    --to=biessmann@corscience.de \
    --cc=ptxdist@pengutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox