From: Per Oberg <pero@wolfram.com>
To: xenomai <xenomai@xenomai.org>
Subject: Re: [Xenomai] Importance of missing modechk/cobolt.wrappers
Date: Thu, 30 Aug 2018 02:17:36 -0500 (CDT) [thread overview]
Message-ID: <1401452424.613986.1535613456917.JavaMail.zimbra@wolfram.com> (raw)
In-Reply-To: <8992f6ad-6627-cb28-32f8-a3e890567e3a@xenomai.org>
----- Den 29 aug 2018, på kl 18:01, Philippe Gerum rpm@xenomai.org skrev:
> On 08/29/2018 05:15 PM, Per Oberg wrote:
> > Hello all
>> I'm working with a Yocto build of xenomai. I have compiled a sdk for building on
>> a build-host but I am doing something wrong because I get the wrong path to
>> modechk.wrappers and cobalt.wrappers. I could probably fix this but I just want
> > to know whether this is a real problem or not.
> > When I run xeno-config i get :
> > Xenomai: wrappers cannot be read from /usr/lib/modechk.wrappers
> > Xenomai: wrappers cannot be read from /usr/lib/cobalt.wrappers
>> The contents of the wrapper files are listed below for completeness. It seems
>> like a lot of stuff that is not getting wrapped when these files are missing,
>> and that would mean that I am missing a lot of wrapping and that I definitely
> > need to fix it. Or is is just a warning?
> It's an important warning, your build may be broken. This means that
> wrapping failed for any executable depending on it, which retrieved its
> linker flags from xeno-config --ldflags. Basically, all built-in Xenomai
> apps (e.g. latency, switchtest), and any user application depending on
> xeno-config for retrieving those flags.
> > How do I know if the final software is wrapping stuff the way I want?
> Dumping the executable's namelist should always refer to the wrapper for
> any routine listed in the wrapper files. e.g.
> $ nm -u $executable | grep 'U __wrap_'
> U __wrap_clock_gettime
> U __wrap_close
> U __wrap_fclose
> U __wrap_fprintf
> U __wrap_fputc
> U __wrap_fputs
> U __wrap_free
> U __wrap_fwrite
> U __wrap_ioctl
> U __wrap_kill
> U __wrap_malloc
> U __wrap_open
> U __wrap_printf
> U __wrap_pthread_attr_init
> U __wrap_pthread_create
> U __wrap_pthread_join
> U __wrap_pthread_setname_np
> U __wrap_puts
> U __wrap_read
> U __wrap_sem_close
> U __wrap_sem_open
> U __wrap_sem_post
> U __wrap_sem_unlink
> U __wrap_sem_wait
> U __wrap_time
> U __wrap_timerfd_create
> U __wrap_timerfd_settime
> U __wrap_write
> --
> Philippe.
Thanks for that, I suspected this was the case.
I did do some checking on the built-in Xenomai apps and they are compiled using the right flags. This does not puzzle me because we are usually able to trust the magic of the build system for solving this. (Meaning: Since there are no libraries, linker files, or other dependencies in the "prefix" directories during the first build the tools must all be referring to the build-dir instead of the prefix destination.)
In my case, since I'm building for a target with no build tools installed (which I assume would be quite normal for a lot of people doing real time stuff) I need to prepare some kind of SDK. For the X86 platform I can install Xenomai on my laptop, and have the exact same setup on the target. But I rather not have it this way. Is there some kind of special flag I can use during build of the SDK or how do people generally go about this?
Perhaps this is more of a Yocto-question or question of build-systems in general but I guess Xenomai-users should be quite good at setting these things up.
Thanks
Per Öberg
next prev parent reply other threads:[~2018-08-30 7:17 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-08-29 15:15 [Xenomai] Importance of missing modechk/cobolt.wrappers Per Oberg
2018-08-29 16:01 ` Philippe Gerum
2018-08-30 7:17 ` Per Oberg [this message]
2018-08-30 7:41 ` Per Oberg
2018-08-31 10:19 ` Philippe Gerum
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=1401452424.613986.1535613456917.JavaMail.zimbra@wolfram.com \
--to=pero@wolfram.com \
--cc=xenomai@xenomai.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.