Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Commits on Jul 12, 2010
  1. @larsclausen

    MMC: jz4740: Rework state handling

    larsclausen authored
    Instead of sleeping inside the threaded irq handler there is now a state
    variable which determines in which action has to be executed next after an irq.
Commits on Jul 11, 2010
  1. @larsclausen

    mmc: jz4740: fix typo

    larsclausen authored
Commits on Jul 10, 2010
  1. jz4740: nand: Enable NAND in static memory bank 1.

    This allows access to the NAND when the kernel is loaded through the USB boot mode.
  2. mtd: cc_ftl: Support partition tables inside CC_FTL block devices.

    The default formatting on the Dingoo A320 does not use a partition table.
    However, if a partition table is added when the Dingoo acts as a USB mass storage device, the China Chip firmware can use it.
  3. mtd: Work around potential deadlock.

    This is a temporary measure for 2.6.34; the problem is fixed in 2.6.35.
  4. mtd: cc_ftl: Fixed "read beyond end of device" check.

    The unit for dev->size is sectors, not bytes.
  5. mtd: cc_ftl: New FTL driver for media players using China Chip firmware.

    Tested on Dingoo A320: the FAT partition on the NAND is mounted successfully.
    Read-only for now.
    Not robust against bad FTL admin data yet: the driver won't crash, but it might return unnecessary I/O errors.
Commits on Jul 9, 2010
  1. jz4740: cpufreq: Fixed selected HCLK divider.

    At most one factor 3 is allowed, so if the CCLK divider is 3 or 6, the HCLK divider cannot be 3 times that.
Commits on Jul 8, 2010
  1. @larsclausen
  2. @larsclausen
  3. @larsclausen
  4. @larsclausen
  5. @larsclausen
  6. @larsclausen
  7. @larsclausen
Commits on Jul 6, 2010
  1. a320_defconfig: Re-enabled serial port.

    The inittab in booboo's rootfs puts a getty on ttyS0.
    If there is no serial port, the getty keeps respawning.
  2. a320: Enable NAND.

    Enabled NAND drivers in the defconfig.
    Added reasonable guesses for the NAND layout: partitions and bad block table.
    Added ECC info to match the oob format used by the version of uC/OS-II that comes preinstalled on the Dingoo A320.
  3. jz4740: nand: Repurpose eccpos to contain the offset of the first ECC…

    … byte in the oob.
    The eccpos array is only 64 bytes long, while 4K page NANDs need 72 bytes of ECC.
    This is a temporary measure until upstream has a solution.
  4. a320_defconfig: Disable serial port and console logging over it.

    Logging over serial port was slowing down the boot a lot.
    Those few people who have a serial port on their Dingoo are perfectly capable of building their own kernel with support for it.
    In fact, probably the only reason to add a serial port is to test your own kernels.
Commits on Jul 5, 2010
  1. @larsclausen
Commits on Jul 4, 2010
  1. jz4740: clock: Use a lower PLL frequency if possible, to save energy.

    The documentation says it can save a few milliwatts.
    The resolution is good enough anyway, so saving some energy is more useful.
  2. @larsclausen
  3. @larsclausen

    MMC: jz4740: Don't spin to long waiting for an irq

    larsclausen authored
    If we spin too long while waiting for an irq, we'll enable the interrupt and put
    the spinning thread to sleep and wake it up again from the irq handler.
    This reduces busy-looping time while keeping performance up.
  4. @larsclausen
  5. @larsclausen

    input: Add pwm beep driver

    larsclausen authored
    A driver for pwm based beepers(pcspeakers)
Commits on Jul 3, 2010
  1. jz4740: cpufreq: Updated author info.

    Removed Ingenic because their code was moved to "clock.c", where they are still credited.
    Added my own name because I made significant changes to this driver.
  2. jz4740: cpufreq: Fix statistics on frequency table changes.

    When the maximum CPU frequency changes, the frequency table is recomputed.
    This was done in the "target" handler, which is called after the "cpufreq_stats" module is notified,
    so the statistics would still reflect the old table.
    The fix was moving the table recomputation into the "verify" handler.
Something went wrong with that request. Please try again.