Fixes a race in the STM USB driver which can lead to the device seeing
only 0-length SETUP transfers. The bug ultimately leads to a failed
USB enumeration. The race condition is described in further detail
below.
The current behavior of the IRQ handler for received OUT transfers is
as such:
- Clear RX_CTR bit (at this point, HW sees that SW has acknowledged
previous SETUP/OUT transfer, and the HW will now accept new STATUS
transfers)
- Call out to the client OutStageCallback (nothing significant here)
- Set EP_RX_CNT back to size of max_packet (was previously set to 0 as
we expect to receive 0-length OUT transfer from host. HW uses this
value to limit the amount of data it can receive)
- Set RX_STAT back to VALID (the HW can now receive new STATUS/OUT
transfers)
The important thing to note here is that even before RX_STAT is set to
VALID in the last step, the HW can still receive and process a new
SETUP transfer as long as RX_CTR is cleared (the spec has some detail
about this under section "Control Transfers" on page 867:
https://www.st.com/resource/en/reference_manual/dm00031936.pdf). The
race will occur if we receive a SETUP transfer after clearing RX_CTR
but before adjusting EP_RX_CNT back to max_packet. In this case, the
IRQ handler will run for the newly received SETUP transfer, but the
transfer will have no data associated with it (the driver ends up
using the previous transfer data which was cached).
We can eliminate the race window by waiting to clear RX_CTR only after
we've reset EP_RX_CNT. In this case the HW will ignore any new SETUP
transfers until after the EP_RX_CNT is reset back to the desired
value.
TL;DR most uses of lib/console.h -> lk/console_cmd.h
Move the part that lets a piece of code somewhere in the system to
define a console command from the actual lib/console api to start an
instance of the console. Move in almost every place the user of the
console command definition to the new header, lk/console_cmd.h which is
always in the include path.
Also remove most uses of testing for WITH_LIB_CONSOLE since you can
almost always just safely define it and then let the linker remove it.
SysTick_Config() tries to set the systick priority to max. Comment out
this line since arch/arm/arm-m/arch.c sets the priority to medium to
compete with the rest of the exceptions in the system.
Most of the warnings are new, such as needing to mark fallthroughs on
cases explicitly. A few are based on signed vs unsigned comparisons.
Disable one warning that was annoying about comparing null to arguments
marked nonnull.
