Re: [Tails-dev] USB installer vs. amd64/686-pae kernels [Was…

Delete this message

Reply to this message
Author: intrigeri
Date:  
To: The Tails public development discussion list
Old-Topics: Re: [Tails-dev] tails next big features testing
Subject: Re: [Tails-dev] USB installer vs. amd64/686-pae kernels [Was: tails next big features testing]
intrigeri wrote (21 Mar 2012 20:30:23 GMT) :
>>     [creator:1160] extlinux not found! Only FAT filesystems will be supported
>>     *** glibc detected *** /usr/bin/python: malloc(): memory corruption: 0x0951d720 ***

[...]
>> I think this issue is related with booting with amd64 kernel.


> Ouch.


Looks like this bug is really a ugly one that makes the software
crash, on amd64 kernel, in each of my tests, in a large and creative
variety of ways:
https://tails.boum.org/todo/usb_install_and_upgrade/todo/#index2h2

It also seems to happen with the 686-pae kernel, while I've not run
enough tests to be certain it also happens constantly there too.

So, given:

  * we're already a bit late wrt. the scheduled "big feature freeze"
    date,
  * while the "add 686-pae and amd64 kernels" feature would be
    a welcome bonus for 0.11, not shipping it with this release would
    be no regression,


I'm tempted to simply remove those two additional kernels
merge the feature/usb_installer branch as is into devel,
be done with it, go ahead with the release schedule,
and postpone further investigation to post-0.11 times.

Another possibility would be to ship these two additional kernels, and
document in some "known issues" section that weird incompatibility.
I dislike this possibility on the grounds it would be a UX and user
support mess.

Thoughts?