| Message ID | 20131017135409.GG22434@titan.lakedaemon.net |
|---|---|
| State | New |
| Headers | show |
Jason Cooper <jason@lakedaemon.net> writes: > Guys, > > The only thing exciting here is the workaround for the mv643xx_eth bug. > Whenever you shut off it's clock (and folks have good reason to do > this), it loses it's mac address. In the happy, ideal boot-from-DT > world, this isn't a problem, since the bootloader will set > local-mac-address in the DT. However, we aren't in that world (yet). > > We've been tossing around different solutions to the problem over the > past year or so, and this one seems a) the least painful, and b) most in > line with what the future will hold. eg, the driver (on reload or > resume) will need to read the mac address from the DT, regardless of how > the mac address got there. > > thx, > > Jason. > > The following changes since commit e1cb367de27ca5c186b0f120c3c10a4a0e8edd2e: > > ARM: kirkwood: Add standby support (2013-10-08 15:41:52 +0000) > > are available in the git repository at: > > git://git.infradead.org/linux-mvebu.git tags/soc-3.13-2 > > for you to fetch changes up to d7f5baeedbdb48d4f5bb2b4f6f56f32b7f2f7aff: > > Documentation: arm/Marvell: clarify Armada SoCs that match 78xx0 pattern (2013-10-08 17:27:59 +0000) > > ---------------------------------------------------------------- > mvebu soc changes for v3.13 (round 2) > > - kirkwood > - remove mbus init, pcie clk init > - retain MAC addr for DT ethernet (work around broken IP) > - docs: clarify Armada SoCs Pulled into next/soc. This had a minor conflict in board-dt.c with Sebastian's clk-of-init branch that's part of next/cleanup. Please double check my conflict resolution (currently in our to-build branch, not yet pushed to for-next) Thanks, Kevin
On 10/18/2013 12:20 AM, Kevin Hilman wrote: > Jason Cooper <jason@lakedaemon.net> writes: > >> Guys, >> >> The only thing exciting here is the workaround for the mv643xx_eth bug. >> Whenever you shut off it's clock (and folks have good reason to do >> this), it loses it's mac address. In the happy, ideal boot-from-DT >> world, this isn't a problem, since the bootloader will set >> local-mac-address in the DT. However, we aren't in that world (yet). >> >> We've been tossing around different solutions to the problem over the >> past year or so, and this one seems a) the least painful, and b) most in >> line with what the future will hold. eg, the driver (on reload or >> resume) will need to read the mac address from the DT, regardless of how >> the mac address got there. >> >> thx, >> >> Jason. >> >> The following changes since commit e1cb367de27ca5c186b0f120c3c10a4a0e8edd2e: >> >> ARM: kirkwood: Add standby support (2013-10-08 15:41:52 +0000) >> >> are available in the git repository at: >> >> git://git.infradead.org/linux-mvebu.git tags/soc-3.13-2 >> >> for you to fetch changes up to d7f5baeedbdb48d4f5bb2b4f6f56f32b7f2f7aff: >> >> Documentation: arm/Marvell: clarify Armada SoCs that match 78xx0 pattern (2013-10-08 17:27:59 +0000) >> >> ---------------------------------------------------------------- >> mvebu soc changes for v3.13 (round 2) >> >> - kirkwood >> - remove mbus init, pcie clk init >> - retain MAC addr for DT ethernet (work around broken IP) >> - docs: clarify Armada SoCs > > Pulled into next/soc. > > This had a minor conflict in board-dt.c with Sebastian's clk-of-init > branch that's part of next/cleanup. Please double check my conflict > resolution (currently in our to-build branch, not yet pushed to for-next) Resolution looks sane to me. While looking at it, I realized that we forgot to remove now obsolete kirkwood_legacy_clk_init. I shove a patch to Jason soon, if it makes it into v3.13 great, if not, it will not break anything and can wait for v3.14. Sebastian
On Thu, Oct 17, 2013 at 03:20:57PM -0700, Kevin Hilman wrote: > Jason Cooper <jason@lakedaemon.net> writes: > > > Guys, > > > > The only thing exciting here is the workaround for the mv643xx_eth bug. > > Whenever you shut off it's clock (and folks have good reason to do > > this), it loses it's mac address. In the happy, ideal boot-from-DT > > world, this isn't a problem, since the bootloader will set > > local-mac-address in the DT. However, we aren't in that world (yet). > > > > We've been tossing around different solutions to the problem over the > > past year or so, and this one seems a) the least painful, and b) most in > > line with what the future will hold. eg, the driver (on reload or > > resume) will need to read the mac address from the DT, regardless of how > > the mac address got there. > > > > thx, > > > > Jason. > > > > The following changes since commit e1cb367de27ca5c186b0f120c3c10a4a0e8edd2e: > > > > ARM: kirkwood: Add standby support (2013-10-08 15:41:52 +0000) > > > > are available in the git repository at: > > > > git://git.infradead.org/linux-mvebu.git tags/soc-3.13-2 > > > > for you to fetch changes up to d7f5baeedbdb48d4f5bb2b4f6f56f32b7f2f7aff: > > > > Documentation: arm/Marvell: clarify Armada SoCs that match 78xx0 pattern (2013-10-08 17:27:59 +0000) > > > > ---------------------------------------------------------------- > > mvebu soc changes for v3.13 (round 2) > > > > - kirkwood > > - remove mbus init, pcie clk init > > - retain MAC addr for DT ethernet (work around broken IP) > > - docs: clarify Armada SoCs > > Pulled into next/soc. > > This had a minor conflict in board-dt.c with Sebastian's clk-of-init > branch that's part of next/cleanup. Please double check my conflict > resolution (currently in our to-build branch, not yet pushed to for-next) resolution in a712104 Merge branch 'next/soc' into for-next looks good. thx, Jason.
Guys, The only thing exciting here is the workaround for the mv643xx_eth bug. Whenever you shut off it's clock (and folks have good reason to do this), it loses it's mac address. In the happy, ideal boot-from-DT world, this isn't a problem, since the bootloader will set local-mac-address in the DT. However, we aren't in that world (yet). We've been tossing around different solutions to the problem over the past year or so, and this one seems a) the least painful, and b) most in line with what the future will hold. eg, the driver (on reload or resume) will need to read the mac address from the DT, regardless of how the mac address got there. thx, Jason. The following changes since commit e1cb367de27ca5c186b0f120c3c10a4a0e8edd2e: ARM: kirkwood: Add standby support (2013-10-08 15:41:52 +0000) are available in the git repository at: git://git.infradead.org/linux-mvebu.git tags/soc-3.13-2 for you to fetch changes up to d7f5baeedbdb48d4f5bb2b4f6f56f32b7f2f7aff: Documentation: arm/Marvell: clarify Armada SoCs that match 78xx0 pattern (2013-10-08 17:27:59 +0000) ---------------------------------------------------------------- mvebu soc changes for v3.13 (round 2) - kirkwood - remove mbus init, pcie clk init - retain MAC addr for DT ethernet (work around broken IP) - docs: clarify Armada SoCs ---------------------------------------------------------------- Ezequiel Garcia (2): ARM: kirkwood: Remove unneeded MBus initialization ARM: kirkwood: Remove unneeded PCIe clock adding Kevin Hilman (1): Documentation: arm/Marvell: clarify Armada SoCs that match 78xx0 pattern Sebastian Hesselbarth (1): ARM: kirkwood: retain MAC address for DT ethernet Documentation/arm/Marvell/README | 1 + arch/arm/mach-kirkwood/board-dt.c | 101 +++++++++++++++++++++++++++++++------- 2 files changed, 84 insertions(+), 18 deletions(-)