Released SDCC 4.5.0 and 4.6.0 miscompile some of our code in the mcs51
register allocator (SDCC bugs 4153 and 4154, fixed in trunk r17025 and
r17039). The first make now downloads the 4.6.0 sources, checks their
sha256, applies the two fixes from sdcc/ and installs the compiler into
toolchain/, named after a checksum of sdcc/, so a changed patch builds a
new one. make CC=sdcc still uses an installed compiler.
The switch is reached through four register functions and the SFR_DATA
registers, so a flat register file behind them is enough to run the table
code on the host. A few addresses are live: writing TBL_CTRL with EXECUTE
performs the VLAN or L2 operation at once and clears the bit, STAT_GET
fetches a counter, L2_TBL_FLUSH_CTRL drops entries. Nothing ever reports
busy, so a polling loop runs once and the tests say nothing about what
happens when hardware stalls; that question was settled elsewhere.
The header states the VLAN and L2 entry layouts on its own, so a test can
hold what rtl837x_port.c writes against a second reading of the same bits
rather than against itself. The first suite does that for the VLAN entry,
the PVID register two ports share, the static multicast and management
entries, the per-port flush and the trunk registers. Two stub headers stand
in for the ones a firmware build generates, so the harness needs no SDCC
build first.
The Makefile has the version number and generates a version.h which
has VERSION_SW define that looks like "v<VERSION>-g<GIT_SHORT_HASH>".
Software version shows at boot in the serial console.
And shown with command: `version`.
It can also be requested via `information.json`.
```
{
"ip_address": "192.168.10.247",
"ip_gateway": "192.168.10.1",
"ip_netmask": "255.255.255.0",
"mac_address": "1c:2a:a3:23:00:02",
"sw_ver": "v0.1.0-gd48235f",
"hw_ver": "SWGT024-V2.0"
}
```
d00f
Rob Emery
João Silva
sergewar
René van Dorst