09:35 <larsc> I wish we had the lower part of RAM mapped at 0x0-0xffff

09:39 <larsc> or having it mapped at 0xffff7fff-0xffffffff would also work

10:12 <lekernel> what for?

10:18 <larsc> it would allow to access memory without loading the address into a register first

10:18 <larsc> using r0 as the base register and a 16 bit signed offset

10:19 <larsc> when we switch from userspace to kernelspace stack in Linux we obviously have to do this without trashing any of the userspace registers

10:19 <larsc> the current code makes a backup of the registers it is going to trash on the userspace stack

10:19 <larsc> but that's a bit ugly

10:36 <lekernel> can't map it with the MMU?

10:38 <larsc> well, this is the no-mmu code

14:15 <GitHub13> [linux-milkymist] larsclausen pushed 7 new commits to master: http://git.io/O2rNcQ

14:15 <GitHub13> linux-milkymist/master a756401 Lars-Peter Clausen: lm32: Implement user_stack_pointer()...

14:15 <GitHub13> linux-milkymist/master b0017d7 Lars-Peter Clausen: Drop __devinit/__devexit from milkymist drivers...

14:15 <GitHub13> linux-milkymist/master 37807cf Lars-Peter Clausen: lm32: Drop unused debugger_info field from thread_info struct...

14:55 <GitHub171> [linux-milkymist] larsclausen pushed 1000 new commits to master: http://git.io/WPJH-A

14:55 <GitHub171> linux-milkymist/master 9d43a18 Michael S. Tsirkin: tun: avoid owner checks on IFF_ATTACH_QUEUE...

14:55 <GitHub171> linux-milkymist/master ec82f94 Lars-Peter Clausen: bfin_mac: Restore hardware time-stamping dependency on BF518...

14:55 <GitHub171> linux-milkymist/master 337da3e Dan Carpenter: bnx2x: move debugging code before the return...

15:02 <larsc> lekernel: how do you keep the hdmi connectors in place? fill the holes with soldering lead?

15:05 <lekernel> yes

15:06 <lekernel> fill just one first and then try to solder the pins, so it's easier to move it again if you mess up

15:07 <lekernel> the TVS are a bit tricky to solder, I've had success by melting the solder on all four pins simultaneously and then plunging the component into the liquid solder

15:08 <lekernel> if you have a reflow oven, you can do that properly ...

15:11 <lekernel> you can check the TVS connections by forward biasing the diode, you should have ~-0.7V between GND and the protected pin

17:10 <roh> lekernel: would hot air help?

17:51 <GitHub157> [migen] sbourdeauducq pushed 3 new commits to master: http://git.io/qoe8mg

17:51 <GitHub157> migen/master 6abac59 Sebastien Bourdeauducq: examples/basic/psync: demonstrate the new features

17:51 <GitHub157> migen/master 7c4e6c3 Sebastien Bourdeauducq: fhdl/verilog: support special lowering and overrides

17:51 <GitHub157> migen/master a878db1 Sebastien Bourdeauducq: genlib: clock domain crossing elements

18:01 <GitHub57> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/K9XKmg

18:01 <GitHub57> migen/master c2d54f4 Sebastien Bourdeauducq: examples/psync: cleanup

18:29 <GitHub77> [mibuild] sbourdeauducq pushed 3 new commits to master: http://git.io/EeEdHA

18:29 <GitHub77> mibuild/master ef83342 Sebastien Bourdeauducq: generic_platform/get_verilog: pass additional args to verilog.convert

18:29 <GitHub77> mibuild/master 56ae0f0 Sebastien Bourdeauducq: xilinx_ise: disable SRL extraction on synchronizers

18:29 <GitHub77> mibuild/master 0321513 Sebastien Bourdeauducq: corelogic -> genlib