[vbox-dev] Patch to support installation on non-Sun/Oracle Solaris hosts
Ramshankar
ramshankar.venkataraman at oracle.com
Thu Oct 8 16:35:15 GMT 2015
Hi,
Sorry for the delayed response but I've managed to find some time to
make a bit of progress with this.
For now, I've added a separate function which gets called if the 'uname
-v' reports one of the 3rd party Solaris variants (OI/Illumos etc.).
That function however just declares it as S11/151. I also added your
suggestion for the vboxinst_vboxusb touch file although I modified your
code a little (checking presence of vboxusbmon.conf first and then
checking the remaining bits).
These changes have been committed to trunk and I intend to backport this
to 5.0.x branch in the next few days. Once this is in, we can add better
support for these distros' kernel easily as risk of breaking the
official Solaris versions are minimized.
If you want an immediate diff, I've attached it here.
Regards,
Ram.
On 09/14/15 05:11 PM, Ramshankar wrote:
> Hi,
>
> Not until the problems are ironed out that all the cases work, or we
> make the patch
> safe for the existing cases (official S10/S11 versions) and have a
> distinct separate
> code path for all the remaining cases.
>
> I'll take a look later this week when I have some time if it's possible
> to separate
> first using uname, and then drill down into the pkg(5) kernel package
> version parsing.
>
> My primary concern is to not break any of the existing cases.
>
> Regards,
> Ram.
>
>
> On 09/11/2015 01:14 PM, Jim Klimov wrote:
>> Hi Ram,
>>
>> Is there any progress about integrating some or all of this
>> vboxconfog.sh improvement into the upstream releases?
>>
>> Thanks,
>> Jim
>>
>> On 2015-07-22 18:48, Jim Klimov wrote:
>>> Hello Ram,
>>>
>>>
>>> Sorry I was away from networking for a while so could not check.
>>>
>>>
>>> I ran vboxconfig.sh replicas modified by your and by my patches,
>>> instrumented to print out major/minor versions after get_sysinfo and
>>> abort. Results differ:
>>>
>>>
>>> OI_151a8 : both detected as 11 / 151
>>>
>>> PKGFMRI=pkg://openindiana.org/system/kernel@0.5.11,5.11-0.151.1.8:20130721T133142Z
>>>
>>>
>>> OmniOS Bloody 151015: mine detected as 11 / 151, yours detected as 11 /
>>> 151015 (which will be greater than 159 in a later test)
>>>
>>> PKGFMRI=pkg://omnios/system/kernel@0.5.11,5.11-0.151015:20150714T040209Z
>>>
>>>
>>> OI Hipster: mine detected as 11 / 151, yours detected as 11 / 0
>>>
>>> PKGFMRI=pkg://openindiana.org/system/kernel@0.5.11-2015.0.1.15147:20150721T130056Z
>>>
>>>
>>> SXCE snv_117 (svr4): mine detected as 11 / 117 (interim
>>> HOST_OS_MAJORVERSION=SXCE), yours detected as a strange Solaris 11
>>> without pkg(5) and aborted... VirtualBox does run there (at least an
>>> older version, and IIRC without both of the advanced modules) ;-)
>>>
>>>
>>> Jim
>>> ----- Исходное сообщение -----
>>> От: Ramshankar <ramshankar.venkataraman at oracle.com>
>>> Дата: Tuesday, July 7, 2015 16:25
>>> Тема: Re: [vbox-dev] Patch to support installation on non-Sun/Oracle
>>> Solaris hosts
>>> Кому (To): jim at cos.ru
>>> Копия (Cc): vbox-dev at virtualbox.org
>>>
>>> > Hi Jim,
>>> >
>>> > On 07/ 7/15 02:45 PM, Jim Klimov wrote:
>>> > > Hello Ram,
>>> > >
>>> > > Makes sense, and that's why I started with touch-files
>>> > following an
>>> > > established example so as to support anything regardless of
>>> > matching,> only an admin (or wrap-package) who'd touch the file
>>> > is needed. ;)
>>> >
>>> > Adding a touch file is okay. I have no objections to having that
>>> > option in.
>>> >
>>> > > I think a "case" would be more readable and conservative on
>>> > resources> than a stack of if-elif-else-fi (don't have to call
>>> > an external program
>>> > > too many times) and would suffice especially since you are comparing
>>> > > fixed strings or, at most, simple wildcards here.
>>> >
>>> > Yes, I'm aware of this but as of now, I'd rather have everything
>>> > working
>>> > than
>>> > fork-avoidance optimizations. This can come at a later step.
>>> >
>>> > > In the patch itself, here's a bit that worries me: either I've
>>> > had a
>>> > > night too sleepless, or there is a logical error here:
>>> > >
>>> > >
>>> > + STR_KERN_MAJOR=`echo "$PKGFMRI" | sed
>>> > > 's/^.*\@//;s/\,.*//;s/\-.*//'`
>>> > >
>>> > + if test "$STR_KERN_MAJOR" = "5.12"; then
>>> > > + ###(comments snipped)
>>> > >
>>> > + BRANCH_VERSION=$STR_KERN_MAJOR
>>> > >
>>> > + HOST_OS_MAJORVERSION=`echo "$BRANCH_VERSION" |
>>> > > cut -f2 -d'-' | cut -f1,2 -d'.'`
>>> > >
>>> > + if test "$HOST_OS_MAJORVERSION" = "5.12"; then
>>> > >
>>> > + HOST_OS_MAJORVERSION="12"
>>> > >
>>> > + HOST_OS_MINORVERSION=`echo
>>> > > "$BRANCH_VERSION" | cut -f2 -d'-' | cut -f6 -d'.'`
>>> > >
>>> > + return 0
>>> > >
>>> > + else
>>> > >
>>> > + errorprint "Failed to parse the Solaris
>>> > > kernel major version."
>>> > >
>>> > + exit 1
>>> > >
>>> > > Here you have STR_KERN_MAJOR=="5.12" (fixed, no more, no
>>> > less), then
>>> > > assign BRANCH_VERSION to the same "5.12" and try to `cut`
>>> > major/minor> versions out of it. You do get "5.12" again as
>>> > major, but minor without
>>> > > the dash-separated part of the FMRI is meaningless (and ends
>>> > up empty).
>>> >
>>> > Oops, my mistake. Thanks. I've attached a new patch of that part
>>> > of the code
>>> > which I think should address the problem.
>>> >
>>> > > Otherwise, cutting the "5.11", "0.5.11" or "5.12" parts seems
>>> > to work
>>> > > correctly here.
>>> > >
>>> > > Also, in line 10 (added comment for OI Hipster string) there
>>> > is "of"
>>> > > instead of "or" ;)
>>> >
>>> > Also fixed in the patch I've attached.
>>> >
>>> > Let me know how this works for the OS/distros in question here.
>>> >
>>> > Regards,
>>> > Ram.
>>> >
>>>
>>>
>>> _______________________________________________
>>> vbox-dev mailing list
>>> vbox-dev at virtualbox.org
>>> https://www.virtualbox.org/mailman/listinfo/vbox-dev
>>>
>> _______________________________________________
>> vbox-dev mailing list
>> vbox-dev at virtualbox.org
>> https://www.virtualbox.org/mailman/listinfo/vbox-dev
>
> _______________________________________________
> vbox-dev mailing list
> vbox-dev at virtualbox.org
> https://www.virtualbox.org/mailman/listinfo/vbox-dev
-------------- next part --------------
A non-text attachment was scrubbed...
Name: changeset_r103214.diff
Type: text/x-patch
Size: 5950 bytes
Desc: not available
URL: <http://www.virtualbox.org/pipermail/vbox-dev/attachments/20151008/f3415e5e/attachment.bin>
More information about the vbox-dev
mailing list