Move ThinkPenguin CCS case study to first one.
The original text for this case study was developed as the final
chapter, as sort of a "you heard about the rest, now try the best"
approach. However, I think we want to put this one first. "Hope before
despair" is perhaps a better attitude?
The diff on this one didn't come out so clean, so it's a bit difficult
to see that I'm just moving a big block of text up.
I diffed it against a version of this file, from:
commit 2f2e5f9e4c on 2014-10-15 22:14:45 -0400
that was before work began on this chapter, and the diff was relatively
clean.
This commit is contained in:
Bradley M. Kuhn
parent
983b035d68
commit
72f50f68ea
1 changed files with 299 additions and 300 deletions
|
|
@ -239,6 +239,305 @@ compliance. We have received the gamut of possible reactions to such
|
|||
requests, and in this course, we examine four specific examples of such
|
||||
compliance work.
|
||||
|
||||
% FIXME: make this section properly TeX-formatted
|
||||
\chapter{ThinkPenguin Wireless Router: A study in Excellent CCS}
|
||||
|
||||
This case study does a step-by-step build and installation analysis of one
|
||||
of the best Complete, Corresponding Source (CCS) releases we've seen. The
|
||||
CSS release studied here was provided for the binary distribution of a
|
||||
physical product by ThinkPenguin. The product is the model
|
||||
``TPE-NWIFIROUTER'', a wireless router.
|
||||
|
||||
The method of
|
||||
distribution (complete source accompanying the product) and the way the source
|
||||
was laid out provide very good examples of how to make things easier for both
|
||||
the distributor and the purchaser of the hardware containing GPLed components.
|
||||
|
||||
\section{Root Filesystem and Kernel Compilation}
|
||||
|
||||
* We found a CD included in the box that the ThinkPenguin TPE-NWIFIROUTER
|
||||
shipped in, labelled "libreCMC v1.2.1 source code". On the CD, there was a
|
||||
README file at the top level, which mentioned that to build the software, one
|
||||
needed a GNU/Linux system as well as a list of approximately 10 packages.
|
||||
These sorts of plain text instructions are helpful because we know what kind
|
||||
of system we are expected to use, and what commands we should run on it. Such
|
||||
instructions are not strictly required, as an obviously-named shell script may
|
||||
suffice, but they are helpful in clarifying any ambiguities that may arise.
|
||||
** Since it appears that this source release will build on a wide range of
|
||||
distributions, it was fine that no specific distribution was specified.
|
||||
However, most source releases we see will only build on a very specific
|
||||
distribution, due to a variety of assumptions made about the build
|
||||
environment. While such a situation is not ideal in the general sense, it is
|
||||
fine to specify a particular distribution that must be use to build the
|
||||
source release (such as "Debian 7 amd64"), from a compliance perspective.
|
||||
As an example, we noticed such an assumption later on in this source release,
|
||||
but it would be easy to correct in the instructions in this situation (see
|
||||
"`GLIBC\verb0_02.14' not found" below).
|
||||
|
||||
% FIXME: Spend some time here (admittedly a digression: maybe refer to
|
||||
% another section later?) about how it's ok to specify a specific build
|
||||
% environment.
|
||||
% FIXME(dg): Hopefully the above will suffice. I can expand more/differently if
|
||||
% such is desired.
|
||||
|
||||
* The actual building of the source code was completed in the following way:
|
||||
** Since the instructions didn't mention a specific distro to use, we ran the
|
||||
build on an amd64 Debian 6 machine we had. The only distro requirement was:
|
||||
|
||||
To build your own firmware you need to have access to a GNU/Linux system
|
||||
(case-sensitive filesystem required).
|
||||
|
||||
** The README mentioned that:
|
||||
|
||||
"In order to build firmware images for your router,the
|
||||
following needs to be installed :
|
||||
|
||||
gcc, binutils, bzip2, flex, python, perl, make, find,
|
||||
grep, diff, unzip, gawk, getopt, libz-dev and libc headers."
|
||||
|
||||
So we ran "dpkg --list" and confirmed that each package was installed (this
|
||||
is indicated by a leading "ii" on the line containing the package). Other
|
||||
GNU/Linux distributions may have other ways of determing which packages are
|
||||
installed.
|
||||
** We then extracted the LibreCMC tarball by running
|
||||
"tar --posix -jxpf /media/libreCMC\verb0_0v1\verb0_02\verb0_01\verb0_0SRC/librecmc-v1.2.1.tar.bz2". The
|
||||
CD did contain another tarball (librecmc-u-boot.tar.bz2), but there appeared
|
||||
to be separate instructions for that (in the u-boot\verb0_0reflash text file in the
|
||||
same directory). Having the README be more explicit about this would be nice
|
||||
but did not ultimately prevent us from determing the proper steps to execute.
|
||||
** The README mentioned the following optional step, which we skipped because
|
||||
we did not need to modify the configuration for our initial build:
|
||||
|
||||
Please use "make menuconfig" to configure your appreciated
|
||||
configuration for the toolchain and firmware. Please note that
|
||||
the default configuration is what was used to build the firmware
|
||||
image for your router. It is advised that you use this configuration.
|
||||
|
||||
** The next instruction was 'Simply running "make" will build your firmware.'
|
||||
So we entered the "librecmc" directory that had been created from the above
|
||||
"tar" command and then ran "make". The build took about 40 minutes to run on
|
||||
our system. The command used and output from running it are available here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0librecmc-complete.log
|
||||
|
||||
% FIXME: Above, I'd like to see more ``walk through'' of the step by step
|
||||
% instructions. The text is a bit terse: could be expanded to talk more.
|
||||
% FIXME(dg): Hopefully the above will suffice. I can expand more/differently if
|
||||
% such is desired.
|
||||
|
||||
* It was helpful to know that we could use "make menuconfig" for configuration
|
||||
changes, as being able to modify the source is an important part of the GPL's
|
||||
requirements. Adding instructions like these shows that the distributor is
|
||||
aware of and interested in promoting the spirit of the GPL, by making it
|
||||
easier to modify the source than may be strictly required by the GPL's text.
|
||||
|
||||
% FIXME: We should somewhere (perhaps on each step we discuss) talk about
|
||||
% what often goes wrong on those steps, and why this is right. As written
|
||||
% now, there is no driving home of the fact that it is uncommon that things
|
||||
% are so smooth. :)
|
||||
% FIXME(dg): Hopefully the below will suffice. I can expand more/differently if
|
||||
% such is desired. (I presume the above comment relates to the below text.)
|
||||
|
||||
* The "make" step completed successfully on our system and resulted in several
|
||||
files being generated in the bin/ar71xx directory, namely firmware images.
|
||||
** This step is normally where we run into the greatest number of build issues
|
||||
(and thus compliance problems). In many cases, the "make" step will fail due
|
||||
to a missing package or because toolchain paths are not setup correctly. As
|
||||
a result, it is important to test the provided instructions on a clean system
|
||||
before distributing the binaries and corresponding source. Listing the
|
||||
specific GNU/Linux distribution and any non-default packages required for the
|
||||
build (ie. those installed before testing the instructions) in the build
|
||||
instructions makes it easier for the end user to successfully build the
|
||||
source release.
|
||||
|
||||
* There appeared to be several filesystem and kernel images, for different
|
||||
hardware versions. It was unclear which one to install on the particular
|
||||
device we received or how to install it, both of which should have been
|
||||
mentioned in the README.
|
||||
|
||||
% FIXME: Below, we probably want to talk to them to add this, and also, be a
|
||||
% bit more expansive.
|
||||
|
||||
* The above installation issue is mitigated by the availability of a web UI in
|
||||
the product that performs firmware image installation. It would be best if
|
||||
instructions like those at http://librecmc.org/librecmc/wiki?name=Tp+MR3020
|
||||
were included in the README, as the user cannot be expected to infer that or
|
||||
to find such a link.
|
||||
|
||||
\section{Root Filesystem and Kernel Installation}
|
||||
|
||||
As mentioned above, the specific steps for installing an updated firmware image
|
||||
were not provided, but we found that the firmware update method available in the
|
||||
web interface worked fine. In particular, we went to http://192.168.10.1/ in
|
||||
our browser, then logged in and chose System -> Backup / Flash Firmware. From
|
||||
there, we went to the "Flash new firmware image" section and selected the
|
||||
librecmc-ar71xx-generic-tl-wr841n-v8-squashfs-sysupgrade.bin image in the
|
||||
bin/ar71xx directory mentioned above. We chose the "v8" image because we found
|
||||
our router said "v8.2" on the bottom and "sysupgrade" because we were doing a
|
||||
firmware upgrade rather than a fresh install.
|
||||
|
||||
When we clicked "Flash image...", we were prompted to confirm the MD5 hash of
|
||||
the image to flash and then clicked "Proceed" to flash the image. The process
|
||||
took about one minute, at which point we were back at the web UI login screen.
|
||||
We logged in and found that the Kernel Log section showed we were running the
|
||||
new kernel.
|
||||
|
||||
We then logged in via SSH again and ran "busybox", which printed the new version
|
||||
string, showing it was using our newly-compiled version (given the date).
|
||||
|
||||
\section{U-Boot Compilation}
|
||||
|
||||
* As mentioned above, we also found a "u-boot\verb0_0reflash" file at the top level of
|
||||
the included source CD. We followed the instructions for compiling U-Boot,
|
||||
which were fairly straight-forward. One modification would be to mention that
|
||||
"\$U-BOOT\verb0_0SRC" referred to the extracted source directory, which was implied,
|
||||
but should have been explicit.
|
||||
* Additionally, we noticed that the included toolchain binaries, which were used
|
||||
by the U-Boot compilation process by default, did not run on our system. In
|
||||
particular, we received this error:
|
||||
|
||||
mips-librecmc-linux-uclibc-gcc.bin: /lib/libc.so.6: version `GLIBC`\verb0_02.14' not found (required by mips-librecmc-linux-uclibc-gcc.bin)
|
||||
|
||||
The complete log output (including the command used to run it) is here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0u-boot-build\verb0_0fail.log
|
||||
|
||||
* We found that by removing toolchain/bin and symlinking the toolchain built for
|
||||
the filesystem/kernel above in its place, we were able to complete the U-Boot
|
||||
build. Specifically, we symlinked toolchain/bin to:
|
||||
|
||||
../../staging\verb0_0dir/toolchain-mips\verb0_034kc\verb0_0gcc-4.6-linaro\verb0_0uClibc-0.9.33.2/bin
|
||||
|
||||
Output from the symlink operation can be found here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0u-boot-create\verb0_0symlink.log
|
||||
|
||||
* Ideally the pre-built toolchain binaries should not be included and a symlink
|
||||
as mentioned above should be created by default, with a mention that the
|
||||
U-Boot build depends on the previous build for its toolchain.
|
||||
* After compilation completed successfully, we found a new U-Boot image in the
|
||||
bin directory. The instructions explained how to install it on the device.
|
||||
Output from the successful build (after the symlink was created) is here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0u-boot-finish\verb0_0build.log
|
||||
|
||||
\section{U-Boot Installation}
|
||||
|
||||
We obtained a serial cable along with our router, in order to complete the
|
||||
U-Boot installation per the instructions in u-boot\verb0_0reflash. However, we were
|
||||
only able to read data from the serial port; we were unable to interrupt the
|
||||
boot process or access the U-Boot console to complete the U-Boot re-flash. Here
|
||||
are the steps we tried:
|
||||
|
||||
* We found the serial cable included was a USB serial adapter that had a male
|
||||
USB type A connector on one end and 4 female jumper wires at the other end.
|
||||
These female jumper wires were red, black, white, and green.
|
||||
* The instructions did not specify how to connect these wires, but we were able
|
||||
to determine this in part using the "v8.4" image (close to our "v8.2" router)
|
||||
at \url{http://wiki.openwrt.org/toh/tp-link/tl-wr841nd#serial.console} . Aside from
|
||||
power and ground (red and black), we did have to guess which of the wires was
|
||||
RX and TX. By experimentation we found that green was RX and white was TX.
|
||||
When we tried the other way, we received no data to our serial console at boot
|
||||
time.
|
||||
* We did have to use the included jumper pin gender changer with the USB serial
|
||||
adapter, which we put through the holes on the router's mainboard and then
|
||||
connected to the USB serial adapter. The fit was fairly loose so it would be
|
||||
nice if future router versions included a tighter gender changer or (ideally)
|
||||
had the jumper pins soldered onto the board to begin with (so no gender
|
||||
changer would be required).
|
||||
* We used 115200 8N1 as our serial console settings (with no hardware or
|
||||
software flow control). This was tested with both the minicom and screen
|
||||
commands. We found that if we connected all 4 wires on the USB serial adapter
|
||||
that the router would start without additional power and our console would
|
||||
receive the startup messages. We could replicate the same behavior by
|
||||
omitting the power cable from the USB serial adapter (red wire) and connecting
|
||||
the main power adapter to the router instead.
|
||||
* While we did see the U-Boot and kernel boot logs in our serial console, we
|
||||
were unable to interrupt the boot process as u-boot\verb0_0reflash indicated we
|
||||
should. We suspect this is a misconfiguration of our serial console, but it's
|
||||
unclear exactly how it is misconfigured, as we were able to receive data fine
|
||||
(we just couldn't send data to the router).
|
||||
* As a result, we were unable to complete the U-Boot installation test. We did
|
||||
appreciate that installation instructions were included, though these
|
||||
instructions should be updated to include more specifics about connecting the
|
||||
serial cable. Since ThinkPenguin does have the option to ship a serial
|
||||
adapter with the router, it would be helpful if instructions specific to that
|
||||
adapter were included, as the wiring configuration one should use was unclear.
|
||||
* Additionally, instructions for removing the router's case should be included.
|
||||
We found that the two screws that needed removal to open the case were hidden
|
||||
underneath rubber feet on the case. Indicating which feet need removal to
|
||||
unscrew the case would be helpful. The instructions should also note that the
|
||||
case needs to be carefully separated once the screws are removed; it
|
||||
effectively snaps apart, but care must be taken to avoid breaking the plastic
|
||||
fasteners that keep the case together after the screws are removed.
|
||||
|
||||
\section{Firmware Comparison}
|
||||
|
||||
To ensure that the CCS did indeed correspond to the firmware that was shipped on
|
||||
the router, we compared the firmware image that we built using the above steps
|
||||
with the filesystem we found on the device itself. The comparison steps we used
|
||||
were:
|
||||
|
||||
* Extract the filesystem from the image we built by running find-firmware.pl
|
||||
from https://gitorious.org/gpl-compliance-tools/gpl-compliance-scripts on
|
||||
librecmc-ar71xx-generic-tl-wr841n-v8-squashfs-factory.bin from the bin/ar71xx
|
||||
directory mentioned above (we noticed that our router said "Ver:8.2" on the
|
||||
bottom). Then run squashfs4.2/squashfs-tools/bat-unsquashfs42 from
|
||||
bat-extratools (at http://www.binaryanalysis.org/en/content/show/download )
|
||||
on the resulting morx0.squash and use the filesystem in the new squashfs-root
|
||||
directory for comparison.
|
||||
* Login to the web interface (at http://192.168.10.1/ ) from a computer that is
|
||||
connected to the router.
|
||||
* Set a password using the provided link at the top (the UI warns that no
|
||||
password is set and asks the user to change it).
|
||||
* Login to the router via SSH, using the root user and the password we just set.
|
||||
* Compare representative directory listings and binaries to ensure the set of
|
||||
included files (on the router) is similar to those found in the firmware image
|
||||
we created (whose contents are now in the local squashfs-root directory). In
|
||||
particular, we did the following comparisons:
|
||||
** List the /bin folder ("ls -l /bin") and confirm the list of files is the same
|
||||
and that the file sizes are similar.
|
||||
** Check the "strings" output of /bin/busybox to confirm it was similar in both
|
||||
places (similar number of lines and content of lines). One cannot directly
|
||||
compare the binaries because the slight compilation variations will cause
|
||||
some bits to be different.
|
||||
** Do the above two steps for /lib/modules, /usr/bin, and other directories with
|
||||
a significant number of binaries.
|
||||
** To check that the kernel is sufficiently similar, compare the "dmesg" output
|
||||
both before and after flashing the new firmware. The kernel version string
|
||||
should be similar, but should have a different build date and user@host
|
||||
indicator. The kernel binary itself is not easily accessible from an SSH
|
||||
login, but may be retrievable using the U-Boot console (the start address of
|
||||
the kernel in flash appears to be 0x9F000000, based on the u-boot\verb0_0reflash
|
||||
instructions). We were not able to verify this, due to the serial connection
|
||||
issues (see above section on U-Boot installation).
|
||||
|
||||
\section{Minor Infractions}
|
||||
|
||||
As mentioned above, there were a few minor infractions. These made it slightly
|
||||
difficult to complete the build and installation without additional context, but
|
||||
did not make the build impossible to complete without more information, such as
|
||||
missing source code for kernel modules or depending on a specific cross-compiler
|
||||
but not mentioning which one or, better yet, including its source code, which
|
||||
are both more problematic infractions. These minor infractions were:
|
||||
|
||||
% FIXME: clarify seriousness of no install instructions; lack of clarity in
|
||||
% which version to install could be more problematic
|
||||
|
||||
* Not mentioning how to extract the source tarball and then where to run the
|
||||
"make" command.
|
||||
* Not mentioning how to install the kernel and root filesystem on the device;
|
||||
this is the biggest of these 3 issues but a bit less troublesome than it would
|
||||
otherwise have been since the web-based firmware update process is well-known.
|
||||
* Using pre-built toolchain binaries that don't work on all systems instead of
|
||||
the ones that are built in a separate step, but not moved to the right place.
|
||||
We were able to build corresponding toolchain binaries from source (though
|
||||
for a slightly different target) so this is not a severe toolchain violation
|
||||
of the type we normally find (where toolchain binaries are provided without
|
||||
source). However, including instructions to use the built toolchain binaries
|
||||
instead would be best, or alternatively specifying the distribution on which
|
||||
the toolchain binaries must be run (to avoid being unable to run them as we
|
||||
were).
|
||||
|
||||
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
||||
\chapter{Bortez: Modified GCC SDK}
|
||||
|
|
@ -799,307 +1098,7 @@ Linux. A decade later, this situation remains largely unresolved.
|
|||
|
||||
\end{enumerate}
|
||||
|
||||
|
||||
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
||||
% FIXME: make this section properly TeX-formatted
|
||||
\chapter{ThinkPenguin Wireless Router: A study in Excellent CCS}
|
||||
|
||||
This case study does a step-by-step build and installation analysis of one
|
||||
of the best Complete, Corresponding Source (CCS) releases we've seen. The
|
||||
CSS release studied here was provided for the binary distribution of a
|
||||
physical product by ThinkPenguin. The product is the model
|
||||
``TPE-NWIFIROUTER'', a wireless router.
|
||||
|
||||
The method of
|
||||
distribution (complete source accompanying the product) and the way the source
|
||||
was laid out provide very good examples of how to make things easier for both
|
||||
the distributor and the purchaser of the hardware containing GPLed components.
|
||||
|
||||
\section{Root Filesystem and Kernel Compilation}
|
||||
|
||||
* We found a CD included in the box that the ThinkPenguin TPE-NWIFIROUTER
|
||||
shipped in, labelled "libreCMC v1.2.1 source code". On the CD, there was a
|
||||
README file at the top level, which mentioned that to build the software, one
|
||||
needed a GNU/Linux system as well as a list of approximately 10 packages.
|
||||
These sorts of plain text instructions are helpful because we know what kind
|
||||
of system we are expected to use, and what commands we should run on it. Such
|
||||
instructions are not strictly required, as an obviously-named shell script may
|
||||
suffice, but they are helpful in clarifying any ambiguities that may arise.
|
||||
** Since it appears that this source release will build on a wide range of
|
||||
distributions, it was fine that no specific distribution was specified.
|
||||
However, most source releases we see will only build on a very specific
|
||||
distribution, due to a variety of assumptions made about the build
|
||||
environment. While such a situation is not ideal in the general sense, it is
|
||||
fine to specify a particular distribution that must be use to build the
|
||||
source release (such as "Debian 7 amd64"), from a compliance perspective.
|
||||
As an example, we noticed such an assumption later on in this source release,
|
||||
but it would be easy to correct in the instructions in this situation (see
|
||||
"`GLIBC\verb0_02.14' not found" below).
|
||||
|
||||
% FIXME: Spend some time here (admittedly a digression: maybe refer to
|
||||
% another section later?) about how it's ok to specify a specific build
|
||||
% environment.
|
||||
% FIXME(dg): Hopefully the above will suffice. I can expand more/differently if
|
||||
% such is desired.
|
||||
|
||||
* The actual building of the source code was completed in the following way:
|
||||
** Since the instructions didn't mention a specific distro to use, we ran the
|
||||
build on an amd64 Debian 6 machine we had. The only distro requirement was:
|
||||
|
||||
To build your own firmware you need to have access to a GNU/Linux system
|
||||
(case-sensitive filesystem required).
|
||||
|
||||
** The README mentioned that:
|
||||
|
||||
"In order to build firmware images for your router,the
|
||||
following needs to be installed :
|
||||
|
||||
gcc, binutils, bzip2, flex, python, perl, make, find,
|
||||
grep, diff, unzip, gawk, getopt, libz-dev and libc headers."
|
||||
|
||||
So we ran "dpkg --list" and confirmed that each package was installed (this
|
||||
is indicated by a leading "ii" on the line containing the package). Other
|
||||
GNU/Linux distributions may have other ways of determing which packages are
|
||||
installed.
|
||||
** We then extracted the LibreCMC tarball by running
|
||||
"tar --posix -jxpf /media/libreCMC\verb0_0v1\verb0_02\verb0_01\verb0_0SRC/librecmc-v1.2.1.tar.bz2". The
|
||||
CD did contain another tarball (librecmc-u-boot.tar.bz2), but there appeared
|
||||
to be separate instructions for that (in the u-boot\verb0_0reflash text file in the
|
||||
same directory). Having the README be more explicit about this would be nice
|
||||
but did not ultimately prevent us from determing the proper steps to execute.
|
||||
** The README mentioned the following optional step, which we skipped because
|
||||
we did not need to modify the configuration for our initial build:
|
||||
|
||||
Please use "make menuconfig" to configure your appreciated
|
||||
configuration for the toolchain and firmware. Please note that
|
||||
the default configuration is what was used to build the firmware
|
||||
image for your router. It is advised that you use this configuration.
|
||||
|
||||
** The next instruction was 'Simply running "make" will build your firmware.'
|
||||
So we entered the "librecmc" directory that had been created from the above
|
||||
"tar" command and then ran "make". The build took about 40 minutes to run on
|
||||
our system. The command used and output from running it are available here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0librecmc-complete.log
|
||||
|
||||
% FIXME: Above, I'd like to see more ``walk through'' of the step by step
|
||||
% instructions. The text is a bit terse: could be expanded to talk more.
|
||||
% FIXME(dg): Hopefully the above will suffice. I can expand more/differently if
|
||||
% such is desired.
|
||||
|
||||
* It was helpful to know that we could use "make menuconfig" for configuration
|
||||
changes, as being able to modify the source is an important part of the GPL's
|
||||
requirements. Adding instructions like these shows that the distributor is
|
||||
aware of and interested in promoting the spirit of the GPL, by making it
|
||||
easier to modify the source than may be strictly required by the GPL's text.
|
||||
|
||||
% FIXME: We should somewhere (perhaps on each step we discuss) talk about
|
||||
% what often goes wrong on those steps, and why this is right. As written
|
||||
% now, there is no driving home of the fact that it is uncommon that things
|
||||
% are so smooth. :)
|
||||
% FIXME(dg): Hopefully the below will suffice. I can expand more/differently if
|
||||
% such is desired. (I presume the above comment relates to the below text.)
|
||||
|
||||
* The "make" step completed successfully on our system and resulted in several
|
||||
files being generated in the bin/ar71xx directory, namely firmware images.
|
||||
** This step is normally where we run into the greatest number of build issues
|
||||
(and thus compliance problems). In many cases, the "make" step will fail due
|
||||
to a missing package or because toolchain paths are not setup correctly. As
|
||||
a result, it is important to test the provided instructions on a clean system
|
||||
before distributing the binaries and corresponding source. Listing the
|
||||
specific GNU/Linux distribution and any non-default packages required for the
|
||||
build (ie. those installed before testing the instructions) in the build
|
||||
instructions makes it easier for the end user to successfully build the
|
||||
source release.
|
||||
|
||||
* There appeared to be several filesystem and kernel images, for different
|
||||
hardware versions. It was unclear which one to install on the particular
|
||||
device we received or how to install it, both of which should have been
|
||||
mentioned in the README.
|
||||
|
||||
% FIXME: Below, we probably want to talk to them to add this, and also, be a
|
||||
% bit more expansive.
|
||||
|
||||
* The above installation issue is mitigated by the availability of a web UI in
|
||||
the product that performs firmware image installation. It would be best if
|
||||
instructions like those at http://librecmc.org/librecmc/wiki?name=Tp+MR3020
|
||||
were included in the README, as the user cannot be expected to infer that or
|
||||
to find such a link.
|
||||
|
||||
\section{Root Filesystem and Kernel Installation}
|
||||
|
||||
As mentioned above, the specific steps for installing an updated firmware image
|
||||
were not provided, but we found that the firmware update method available in the
|
||||
web interface worked fine. In particular, we went to http://192.168.10.1/ in
|
||||
our browser, then logged in and chose System -> Backup / Flash Firmware. From
|
||||
there, we went to the "Flash new firmware image" section and selected the
|
||||
librecmc-ar71xx-generic-tl-wr841n-v8-squashfs-sysupgrade.bin image in the
|
||||
bin/ar71xx directory mentioned above. We chose the "v8" image because we found
|
||||
our router said "v8.2" on the bottom and "sysupgrade" because we were doing a
|
||||
firmware upgrade rather than a fresh install.
|
||||
|
||||
When we clicked "Flash image...", we were prompted to confirm the MD5 hash of
|
||||
the image to flash and then clicked "Proceed" to flash the image. The process
|
||||
took about one minute, at which point we were back at the web UI login screen.
|
||||
We logged in and found that the Kernel Log section showed we were running the
|
||||
new kernel.
|
||||
|
||||
We then logged in via SSH again and ran "busybox", which printed the new version
|
||||
string, showing it was using our newly-compiled version (given the date).
|
||||
|
||||
\section{U-Boot Compilation}
|
||||
|
||||
* As mentioned above, we also found a "u-boot\verb0_0reflash" file at the top level of
|
||||
the included source CD. We followed the instructions for compiling U-Boot,
|
||||
which were fairly straight-forward. One modification would be to mention that
|
||||
"\$U-BOOT\verb0_0SRC" referred to the extracted source directory, which was implied,
|
||||
but should have been explicit.
|
||||
* Additionally, we noticed that the included toolchain binaries, which were used
|
||||
by the U-Boot compilation process by default, did not run on our system. In
|
||||
particular, we received this error:
|
||||
|
||||
mips-librecmc-linux-uclibc-gcc.bin: /lib/libc.so.6: version `GLIBC`\verb0_02.14' not found (required by mips-librecmc-linux-uclibc-gcc.bin)
|
||||
|
||||
The complete log output (including the command used to run it) is here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0u-boot-build\verb0_0fail.log
|
||||
|
||||
* We found that by removing toolchain/bin and symlinking the toolchain built for
|
||||
the filesystem/kernel above in its place, we were able to complete the U-Boot
|
||||
build. Specifically, we symlinked toolchain/bin to:
|
||||
|
||||
../../staging\verb0_0dir/toolchain-mips\verb0_034kc\verb0_0gcc-4.6-linaro\verb0_0uClibc-0.9.33.2/bin
|
||||
|
||||
Output from the symlink operation can be found here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0u-boot-create\verb0_0symlink.log
|
||||
|
||||
* Ideally the pre-built toolchain binaries should not be included and a symlink
|
||||
as mentioned above should be created by default, with a mention that the
|
||||
U-Boot build depends on the previous build for its toolchain.
|
||||
* After compilation completed successfully, we found a new U-Boot image in the
|
||||
bin directory. The instructions explained how to install it on the device.
|
||||
Output from the successful build (after the symlink was created) is here:
|
||||
|
||||
enforcement-case-studies\verb0_0log-output/thinkpenguin\verb0_0u-boot-finish\verb0_0build.log
|
||||
|
||||
\section{U-Boot Installation}
|
||||
|
||||
We obtained a serial cable along with our router, in order to complete the
|
||||
U-Boot installation per the instructions in u-boot\verb0_0reflash. However, we were
|
||||
only able to read data from the serial port; we were unable to interrupt the
|
||||
boot process or access the U-Boot console to complete the U-Boot re-flash. Here
|
||||
are the steps we tried:
|
||||
|
||||
* We found the serial cable included was a USB serial adapter that had a male
|
||||
USB type A connector on one end and 4 female jumper wires at the other end.
|
||||
These female jumper wires were red, black, white, and green.
|
||||
* The instructions did not specify how to connect these wires, but we were able
|
||||
to determine this in part using the "v8.4" image (close to our "v8.2" router)
|
||||
at \url{http://wiki.openwrt.org/toh/tp-link/tl-wr841nd#serial.console} . Aside from
|
||||
power and ground (red and black), we did have to guess which of the wires was
|
||||
RX and TX. By experimentation we found that green was RX and white was TX.
|
||||
When we tried the other way, we received no data to our serial console at boot
|
||||
time.
|
||||
* We did have to use the included jumper pin gender changer with the USB serial
|
||||
adapter, which we put through the holes on the router's mainboard and then
|
||||
connected to the USB serial adapter. The fit was fairly loose so it would be
|
||||
nice if future router versions included a tighter gender changer or (ideally)
|
||||
had the jumper pins soldered onto the board to begin with (so no gender
|
||||
changer would be required).
|
||||
* We used 115200 8N1 as our serial console settings (with no hardware or
|
||||
software flow control). This was tested with both the minicom and screen
|
||||
commands. We found that if we connected all 4 wires on the USB serial adapter
|
||||
that the router would start without additional power and our console would
|
||||
receive the startup messages. We could replicate the same behavior by
|
||||
omitting the power cable from the USB serial adapter (red wire) and connecting
|
||||
the main power adapter to the router instead.
|
||||
* While we did see the U-Boot and kernel boot logs in our serial console, we
|
||||
were unable to interrupt the boot process as u-boot\verb0_0reflash indicated we
|
||||
should. We suspect this is a misconfiguration of our serial console, but it's
|
||||
unclear exactly how it is misconfigured, as we were able to receive data fine
|
||||
(we just couldn't send data to the router).
|
||||
* As a result, we were unable to complete the U-Boot installation test. We did
|
||||
appreciate that installation instructions were included, though these
|
||||
instructions should be updated to include more specifics about connecting the
|
||||
serial cable. Since ThinkPenguin does have the option to ship a serial
|
||||
adapter with the router, it would be helpful if instructions specific to that
|
||||
adapter were included, as the wiring configuration one should use was unclear.
|
||||
* Additionally, instructions for removing the router's case should be included.
|
||||
We found that the two screws that needed removal to open the case were hidden
|
||||
underneath rubber feet on the case. Indicating which feet need removal to
|
||||
unscrew the case would be helpful. The instructions should also note that the
|
||||
case needs to be carefully separated once the screws are removed; it
|
||||
effectively snaps apart, but care must be taken to avoid breaking the plastic
|
||||
fasteners that keep the case together after the screws are removed.
|
||||
|
||||
\section{Firmware Comparison}
|
||||
|
||||
To ensure that the CCS did indeed correspond to the firmware that was shipped on
|
||||
the router, we compared the firmware image that we built using the above steps
|
||||
with the filesystem we found on the device itself. The comparison steps we used
|
||||
were:
|
||||
|
||||
* Extract the filesystem from the image we built by running find-firmware.pl
|
||||
from https://gitorious.org/gpl-compliance-tools/gpl-compliance-scripts on
|
||||
librecmc-ar71xx-generic-tl-wr841n-v8-squashfs-factory.bin from the bin/ar71xx
|
||||
directory mentioned above (we noticed that our router said "Ver:8.2" on the
|
||||
bottom). Then run squashfs4.2/squashfs-tools/bat-unsquashfs42 from
|
||||
bat-extratools (at http://www.binaryanalysis.org/en/content/show/download )
|
||||
on the resulting morx0.squash and use the filesystem in the new squashfs-root
|
||||
directory for comparison.
|
||||
* Login to the web interface (at http://192.168.10.1/ ) from a computer that is
|
||||
connected to the router.
|
||||
* Set a password using the provided link at the top (the UI warns that no
|
||||
password is set and asks the user to change it).
|
||||
* Login to the router via SSH, using the root user and the password we just set.
|
||||
* Compare representative directory listings and binaries to ensure the set of
|
||||
included files (on the router) is similar to those found in the firmware image
|
||||
we created (whose contents are now in the local squashfs-root directory). In
|
||||
particular, we did the following comparisons:
|
||||
** List the /bin folder ("ls -l /bin") and confirm the list of files is the same
|
||||
and that the file sizes are similar.
|
||||
** Check the "strings" output of /bin/busybox to confirm it was similar in both
|
||||
places (similar number of lines and content of lines). One cannot directly
|
||||
compare the binaries because the slight compilation variations will cause
|
||||
some bits to be different.
|
||||
** Do the above two steps for /lib/modules, /usr/bin, and other directories with
|
||||
a significant number of binaries.
|
||||
** To check that the kernel is sufficiently similar, compare the "dmesg" output
|
||||
both before and after flashing the new firmware. The kernel version string
|
||||
should be similar, but should have a different build date and user@host
|
||||
indicator. The kernel binary itself is not easily accessible from an SSH
|
||||
login, but may be retrievable using the U-Boot console (the start address of
|
||||
the kernel in flash appears to be 0x9F000000, based on the u-boot\verb0_0reflash
|
||||
instructions). We were not able to verify this, due to the serial connection
|
||||
issues (see above section on U-Boot installation).
|
||||
|
||||
\section{Minor Infractions}
|
||||
|
||||
As mentioned above, there were a few minor infractions. These made it slightly
|
||||
difficult to complete the build and installation without additional context, but
|
||||
did not make the build impossible to complete without more information, such as
|
||||
missing source code for kernel modules or depending on a specific cross-compiler
|
||||
but not mentioning which one or, better yet, including its source code, which
|
||||
are both more problematic infractions. These minor infractions were:
|
||||
|
||||
% FIXME: clarify seriousness of no install instructions; lack of clarity in
|
||||
% which version to install could be more problematic
|
||||
|
||||
* Not mentioning how to extract the source tarball and then where to run the
|
||||
"make" command.
|
||||
* Not mentioning how to install the kernel and root filesystem on the device;
|
||||
this is the biggest of these 3 issues but a bit less troublesome than it would
|
||||
otherwise have been since the web-based firmware update process is well-known.
|
||||
* Using pre-built toolchain binaries that don't work on all systems instead of
|
||||
the ones that are built in a separate step, but not moved to the right place.
|
||||
We were able to build corresponding toolchain binaries from source (though
|
||||
for a slightly different target) so this is not a severe toolchain violation
|
||||
of the type we normally find (where toolchain binaries are provided without
|
||||
source). However, including instructions to use the built toolchain binaries
|
||||
instead would be best, or alternatively specifying the distribution on which
|
||||
the toolchain binaries must be run (to avoid being unable to run them as we
|
||||
were).
|
||||
|
||||
|
||||
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue