[Lenny400] Patches for linux-stable

H. Nikolaus Schaller hns at goldelico.com
Sat Sep 2 21:25:37 CEST 2017


Hi,

> Am 02.09.2017 um 20:49 schrieb Paul Boddie <paul at boddie.org.uk>:
> 
> On Saturday 2. September 2017 17.01.31 H. Nikolaus Schaller wrote:
>> 
>> I forgot to mention that I added some stubs for i2c and i2s to the
>> jz4730.dtsi.
> 
> I saw this. In fact, I've just added a definition for the DMA controller to 
> the device tree file in my own repository, along with modifications to the 
> driver.

Fine. Something to merge in the future...

> 
>> And of course I added individual patches for keyboard, buttons, leds,
>> sound, etc. to the mipsbook_400.dts. Still untested but should not to be
>> too far away from what we need.
>> 
>> What I did not do is to remove anything from the board file. This needs to
>> be done carefully, including a comparison with the DT so that we do not
>> loose anything important (although we always can look into the git
>> history).
> 
> Indeed.
> 
> Currently, I'm trying to clone the gta04-kernel repository. The first attempt 
> gave me the error I always seem to get with git these days:
> 
> $ git clone git://git.goldelico.com/gta04-kernel.git
> Cloning into 'gta04-kernel'...
> remote: ZÀhle Objekte: 6749163, Fertig.
> remote: Komprimiere Objekte: 100% (976488/976488), Fertig.
> Receiving objects: 100% (6749163/6749163), 1.41 GiB | 703.00 KiB/s, done.
> fatal: pack is corrupted (SHA1 mismatch)

oops.

> fatal: index-pack failed
> remote: Total 6749163 (delta 5725996), reused 6747225 (delta 5724298)
> 
> There is some "advice" about this:
> 
> https://stackoverflow.com/questions/21277806/fatal-early-eof-fatal-index-pack-
> failed
> 
> https://stackoverflow.com/questions/1943914/git-clone-fails-with-index-pack-
> failed
> 
> The consensus seems to involve increasing memory limits. Predictably, there 
> are all sorts of other "remedies" that rather seem like pressing all available 
> buttons and hoping one of them has an effect. Or blaming the network.
> 
> The second attempt got me a clone, even though it also complained:
> 
> $ git clone git://git.goldelico.com/gta04-kernel.git
> Cloning into 'gta04-kernel'...
> remote: ZÀhle Objekte: 6749163, Fertig.
> remote: Komprimiere Objekte: 100% (976488/976488), Fertig.
> Receiving objects: 100% (6749163/6749163), 1.41 GiB | 819 KiB/s, done.
> remote: Total 6749163 (delta 5725996), reused 6747225 (delta 5724298)
> Resolving deltas: 100% (5725996/5725996), done.
> error: unable to find 72df2e1b88afe209c8e9420a28f46eeb865b987a
> error: unable to read sha1 file of drivers/tty/serial/sunsu.c 
> (72df2e1b88afe209c8e9420a28f46eeb865b987a)
> Checking out files: 100% (60075/60075), done.
> 
> Running git fsck produces...
> 
> $ git fsck
> Checking object directories: 100% (256/256), done.
> error: Packfile index for .git/objects/pack/pack-
> c1fd5214844472b9d4a4ba1badfa7f893bbbbfcb.pack SHA1 mismatch
> error: .git/objects/pack/pack-c1fd5214844472b9d4a4ba1badfa7f893bbbbfcb.pack 
> SHA1 checksum mismatch
> error: packed 027e1f5660121287ea7a06f2bfc92f47db93d9ce from 
> .git/objects/pack/pack-c1fd5214844472b9d4a4ba1badfa7f893bbbbfcb.pack is 
> corrupt
> 
> ...and lots of similar messages to the last one. Maybe the clone is still 
> usable, though.

Maybe the repository on the server lost some file (although it should not
be possible to happen due to journaled file system). I will run a git fsck
there.

Or you could try

	git clone http://git.goldelico.com/gta04-kernel.git

Or even the github mirror

	git at github.com:goldelico/gta04-kernel.git


> 
> [...]
> 
>> I now have fixed my cross-gcc includes and it seems to compile fine. I get
>> a 4.2 MB large uImage and 75 MByte kernel modules (unstripped)...
>> 
>> This might be too big for NAND, but for SD card it should be ok.
> 
> I've not focused on the NAND at all so far. Replacing the NAND kernel is 
> probably an "advanced" operation for people who can recover easily in case it 
> goes wrong.

Indeed. I don't see it as high priority. Especially since it is easy to brick
the device by overwriting U-Boot. That is what I had started the L400-JTAG project
long time ago but never finished. To be able to reflash the complete NAND in
case of such a malheur.

> 
> [...]
> 
>> BR and thanks for stimulating the fun while working on this :)
> 
> I hope the fun lasts all the way to a working kernel, though. :-)

Well, the way to get something working for the first time is always fun.
Like climbing a mountain for the first time.

The way to get it into kernel.org is the opposite :) Like climbing the same
mountain over and over even in adverse weather and with a stopwatch in your
neck...

BR,
Nikolaus



More information about the Lenny400 mailing list