Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This is a merge of all previous work done on sdbfs using a different
repository (which I used to push to gitorious).

Besides the kernel stuff, which is still in flush, userspace and lib
are actually useful to sdb users in general, so this is a better place
to host it.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

According to the kernel headers you have, compiling the kernel part
may fail. So compile userspace first, in case you need gensdbfs alone.
Thanks to Matthieu Cattin for noting this.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

This a major thing, that changes a lot. However, now directories work
properly and the code is generally cleaner. This is what I get,
for example, for the wr-nic release gateware on a SPEC:

  spusa.root# find /mnt
  /mnt
  /mnt/WB4-Crossbar-GSI
  /mnt/WR-CORE
  /mnt/WR-CORE/WB4-Crossbar-GSI
  /mnt/WR-CORE/WB4-BlockRAM
  /mnt/WR-CORE/WB4-Bridge-GSI
  /mnt/WR-CORE/WB4-Bridge-GSI/WB4-Crossbar-GSI
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-Mini-NIC
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-Endpoint
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-Soft-PLL
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-PPS-Generator
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-Periph-Syscon
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-Periph-UART
  /mnt/WR-CORE/WB4-Bridge-GSI/WR-Periph-1Wire
  /mnt/WR-CORE/WB4-Bridge-GSI/WB4-Streaming-DMA_0
  /mnt/WR-NIC
  /mnt/WB-VIC-Int.Control
  /mnt/WR-TXTSU
  /mnt/WR-DIO-Core
  /mnt/WR-DIO-Core/WB4-Crossbar-GSI
  /mnt/WR-DIO-Core/WR-1Wire-master
  /mnt/WR-DIO-Core/WB-I2C-Master
  /mnt/WR-DIO-Core/WB-GPIO-Port
  /mnt/WR-DIO-Core/WR-DIO-Registers

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

If the host is little-endian and the bus is directly connected as 32
bits, byte access would be wrong.  This commit detects the thing from
the magic number value and reversed the words of the SDB records.
Missing that, strings would be reversed and 16-bit fields would be in
the wrong byte offset.  Please note that the resulting SDB record
still has big-endian fields.

This problem results from bit 0 being mapped to bit 0 and bit 31 being
mapped to bit 31 (i.e., 32-bit words are native mapped on both sides
of the bridge).  However, on the sdb bus bit 31 (MSB) lives at byte 0,
while the little-endian host stores it at byte 3.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

There is some vestigial code for kernel space and freestanding users, but
those issues will be implemented in later commits.

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>

Signed-off-by: Alessandro Rubini <rubini@gnudd.com>