**** BEGIN LOGGING AT Sat May 11 02:59:57 2019
May 11 07:29:34 <jeffsf> Is there anything in a "default" `master` build that supports file transfer over serial?
May 11 07:30:24 <jeffsf> Not critical, as this device has a great U-Boot implementation
May 11 07:41:45 <jeffsf> (Though the Realtek NIC in the USB-C hub attached to my laptop is apparently a piece of crap)
May 11 09:29:46 <Tapper> I tryed out diizzys patch to change CPU_SUBTYPE:=neon.  but it will not build
May 11 09:30:08 <Tapper> any one here use CPU_SUBTYPE:=neon.  in there builds?
May 11 09:30:54 <Tapper> target  is mvebu
May 11 09:31:33 <Tapper> O I think it's gcc 8 that mite be crashing my build
May 11 09:35:14 <Tapper> Lets see if it was binutils 2.32 that was messing up.
May 11 09:41:57 <ldir> Likes nbd’s kernel testing version patch
May 11 09:44:22 <Tapper> ldir do you know if building with gcc 8 is broken?
May 11 09:45:13 <Tapper> Or if I have to install any extra packages on the OS I am building on?
May 11 09:57:59 <ldir> nothing that I know of
May 11 11:29:44 <ynezz> gch981213: looks nice
May 11 14:52:14 <brucethompson> I have a question about the OpenWrt makefiles
May 11 14:52:49 <brucethompson> I am building a package where I need to not use some of the standard gcc options for OpenWrt
May 11 14:53:15 <brucethompson> The option I need to remove is "-fno-plt"
May 11 14:53:45 <brucethompson> I found the option in rules.mk and as a test I removed it from rules.mk
May 11 14:53:50 <karlp> what the _hell_ is your upstream project doing?!
May 11 14:54:05 <brucethompson> It needs symbols
May 11 14:54:41 <brucethompson> Its a python package
May 11 14:56:28 <brucethompson> The package is PyNacl
May 11 14:57:01 <brucethompson> It builds it own libsodium. It needs the external symbols preserved in that version of libsodium
May 11 14:58:47 <brucethompson> As far as I can tell, "-fno-plt" removes the Program Load Table in the resulting .o file.
May 11 14:59:00 <brucethompson> So I think this is teh option causing the issue
May 11 14:59:38 <brucethompson> When I look at the .o generated by the default OpenWrt Makefile, all the external symbols are undefined.
May 11 15:00:25 <brucethompson> The version of libsodium created for PyNacl needs to have the external symbols correctly defined.
May 11 15:00:42 <brucethompson> PyNacl is a standard python package.
May 11 15:02:20 <brucethompson> <karlp>Still ther??
May 11 15:06:59 <karlp> "standard"
May 11 15:07:10 <karlp> "correctly defined" :)
May 11 15:07:33 <brucethompson> Oh....
May 11 15:07:42 <brucethompson> Bad language
May 11 15:08:48 <brucethompson> I did not write the PyNacl package. I am only trying to port it to OpenWrt
May 11 15:10:36 <karlp> You can completely ovewrite the Compile step, you just hhhave to decide how far you want to go...
May 11 15:11:16 <brucethompson> What I don't understand is where these options are coming from.
May 11 15:11:28 <brucethompson> Doesn't seem like they are coming from rules.mk
May 11 15:12:08 <brucethompson> I checked all the *.mk in the include directory. Doesn;t seem to be coming from there either
May 11 15:21:16 <karlp> probably from the target.
May 11 15:26:58 <brucethompson> You mean the target Makefile?
May 11 15:30:48 <brucethompson> Target is compile
May 11 16:08:19 <lach1012> hmpf, feed failed
May 11 18:22:46 <jeffsf_> ath79: For recent SoCs (those with NAND), is it reasonable to assume that PCI_AR71XX and PHY_AR7100_USB are dead with the advent of PCI_AR724X and PHY_AR7200_USB
May 11 18:23:55 <jeffsf> (QCA956x.dtsi uses the latter ones)
May 11 18:29:18 <m4t> jeffsf: regarding serial transfer, no, but this might work: https://raw.githubusercontent.com/mateusza/shellscripthttpd/master/base64.sh
May 11 18:29:35 <m4t> i just tested and this worked on my router: gzip -c < /sbin/procd | ./base64.sh | ./base64.sh -d |gzip -dc > procd; md5sum procd /sbin/procd
May 11 18:29:48 <m4t> slow though ;)
May 11 18:30:27 <jeffsf> m4t: Thanks!!! That looks like a great one for the emergency tool bag
May 11 18:30:41 <m4t> i've used it before in a pinch
May 11 18:30:52 <m4t> there are some other ones floating around but might not work with busybox awk
May 11 18:31:25 <jeffsf> Yeah, I'm on travel and that POS Ethernet adapter went out on me, stopping me dead
May 11 18:31:35 <m4t> :(
May 11 18:32:13 <jeffsf> It works for a while, then it just stops. Unplug, come back after lunch, and it works again...
May 11 18:32:34 <jeffsf> Three routers with me, but only one dongle
May 11 19:19:41 <mangix> stintel: ping
May 11 19:36:57 <jeffsf> config TESTING_KERNEL -- sweet!!!
May 11 19:37:21 <jeffsf> Thanks nbd!
May 11 19:55:10 <jeffsf> (hmmm, now to figure out enough of the dependency scheme to make ath79-nand dependent on TESTING_KERNEL)
May 11 19:59:27 <KanjiMonster> jeffsf: I don't think you can, but subtargets can define their own kernel versions
May 11 20:00:19 <jeffsf> Ah! OK, that will handle it, thanks!
May 11 20:02:45 <jeffsf> Still worth wading through that merge though
May 11 20:22:25 <jeffsf> KanjiMonster: That would go in `target.mk`, yes?
May 11 20:22:43 <KanjiMonster> jeffsf: right
May 11 20:39:21 <jeffsf> Thanks -- after remembering to rm -rf ./tmp/ it worked like a charm -- neatly solves several bits of ugliness in prepping a patch set
May 12 02:31:11 <gch981213> jeffsf_: Those ar71xx/ar7100 drivers are for ar7130/ar7161/ar913x.
**** ENDING LOGGING AT Sun May 12 03:00:31 2019