Re: [Tails-dev] Please test, review'n'merge feature/sdio

Supprimer ce message

Répondre à ce message
Auteur: intrigeri
Date:  
À: The Tails public development discussion list
Sujet: Re: [Tails-dev] Please test, review'n'merge feature/sdio
Hi,

First, thanks a lot for testing!

sajolida@??? wrote (13 Oct 2013 11:57:16 GMT) :
> - In the menu the launcher is called "Tails installer". It should be
> "Tails Installer" as we use title capitalization for application name,
> as all the other menus do.


Fixed in Git yesterday, so we're on the same page :)

> - In the confirmation dialog, the device is presented as
> "8.0 GB SD8GB device" (note the double space), whereas on the X200 it
> is presented with a more explicit brand name. But that's no big deal.


This discrepancy shows that the reader is filtering what brand / model
the OS can see. I don't think we can do anything about it. Anyway,
that's still better than what we have in 0.20.1, IMHO.

> Apart from that the process when all-right. But then I realized that the
> X61 cannot boot from SD cards. So that's a bit useless...


Sorry about that :(

> 2. So I tried to boot that SD card on a ThinkPad X200 and it wouldn't
> boot. The syslinux error message was:


>     SYSLINUX 4.02 debian-20101014 EDD Load error - Boot error


Hmmm. I'm now offline so I can't check, but this syslinux error
reminds me of USB sticks that I've seen with a MBR/GPT in weird state.
Usually, this was fixed by blanking the first few MB of this SD card,
and re-installing. Could you please try this?

Also, I suspect the MBR or GPT wasn't properly sync'd on the SD card,
so please use GNOME Disk Utility to "safely remove" it after an
apparently successful installation.

> 3. On the X200, I tried to run Tails Installer from this same branch and
> clone onto the same SD card, boot from the X200, and I got the same
> error message. I'm sending you in a private email most of the files on
> that SD card for debugging.


Thanks, I'll have a look tomorrow (I doubt there will be anything
useful in there, but still). Hopefully I won't have to ask you to
provide a copy of the first few MB of the block device, as I'm not too
thrilled to dig into such a dump in the next few days.

> 4. Then on the X200, I tried to run Tails Installer from 0.21 and clone


You really mean 0.20.1, here, I guess. Ignore me if I guessed right.

> onto that SD card, boot from the X200, and that worked fine. This worked
> all-right with the feature/liveusb_ui_improvement that I tested on
> Wednesday as well.


This is useful info, that should allow us to bisect the problem
more easily.

> 5. If I clone onto a USB stick using this same branch from the X61, the
> USB stick boots fine.


Does "this same branch" here mean feature/sdio or
feature/liveusb_ui_improvement?

If it means feature/sdio, then this is *very* weird: there's no
difference in how the installation process is done, between USB
and sdio.

> So, from my experience, feature/sdio, adds a regression that prevents
> from installing onto a SD card either through USB or through SD, and
> boot through USB.


Oh crap, and the freeze is in 4 days. Interesting. I haven't seen any
such problem during my own tests, and 99% of the code changes are in
the UI and some refactoring.

> What else can I do to help your track this down?


I can't reproduce this, so I think we have to resort to Git bisect,
and you'll have to run each step.

Unfortunately, IIRC some of the commits on this branch are buggy, so
that won't be trivial. Either I could provide you with a set of
liveusb-creator .deb's that you could use to bisect the problem more
easily. Or I could prepare a cleaned-up version of the branch, that
you can use to have a funny `git bisect' session yourself. (I tend to
prefer the second solution, but that's your call.)

On my side, I'll see if I can get a hand of a X61.

Cheers,
--
intrigeri
| GnuPG key @ https://gaffer.ptitcanardnoir.org/intrigeri/intrigeri.asc
| OTR fingerprint @ https://gaffer.ptitcanardnoir.org/intrigeri/otr.asc