Re: [PATCH net 2/2] net: usb: lan78xx: reject short register and statistics reads
From: Wei Jie Law
Date: Sun Oct 18 2026 - 02:16:12 EST
Sashiko is correct. Will be submitting v2 with commit messages changed:
- with patch 1 a zero-length stats reply is already harmless; the remaining
leak is a 1 to 187 byte reply (the counter block is 188 bytes). The message
now says so and no longer claims the counters "stay zero" because of 2/2.
- it says that 2/2 depends on 1/2 (otherwise the leak moves to the stack) and
that both should be backported together.
- RX_ADDRL/RX_ADDRH are only written back for an invalid address; the heap
derived address is written to MAF_LO(0)/MAF_HI(0) and set as the netdev address.
pw-bot: cr