On 12/05/2015 02:58 AM, Andrew Baumann wrote: > There appears to have been a longstanding typo in the implementation > of the "MAC address loaded" bit in the E2P_CMD (EEPROM command) > register. The code was using 0x10, but the controller spec says it > should be bit 8 (0x100). > > Signed-off-by: Andrew Baumann <andrew.baum...@microsoft.com> > --- > This may have slipped through, because the Linux driver doesn't > check that bit; it simply reads the MAC address and assumes it is > valid.
Applied in my -net for 2.5. Thanks > > hw/net/lan9118.c | 8 +++++--- > 1 file changed, 5 insertions(+), 3 deletions(-) > > diff --git a/hw/net/lan9118.c b/hw/net/lan9118.c > index 4f0e840..133fd3d 100644 > --- a/hw/net/lan9118.c > +++ b/hw/net/lan9118.c > @@ -56,6 +56,8 @@ do { fprintf(stderr, "lan9118: error: " fmt , ## > __VA_ARGS__);} while (0) > #define CSR_E2P_CMD 0xb0 > #define CSR_E2P_DATA 0xb4 > > +#define E2P_CMD_MAC_ADDR_LOADED 0x100 > + > /* IRQ_CFG */ > #define IRQ_INT 0x00001000 > #define IRQ_EN 0x00000100 > @@ -352,14 +354,14 @@ static void lan9118_reload_eeprom(lan9118_state *s) > { > int i; > if (s->eeprom[0] != 0xa5) { > - s->e2p_cmd &= ~0x10; > + s->e2p_cmd &= ~E2P_CMD_MAC_ADDR_LOADED; > DPRINTF("MACADDR load failed\n"); > return; > } > for (i = 0; i < 6; i++) { > s->conf.macaddr.a[i] = s->eeprom[i + 1]; > } > - s->e2p_cmd |= 0x10; > + s->e2p_cmd |= E2P_CMD_MAC_ADDR_LOADED; > DPRINTF("MACADDR loaded from eeprom\n"); > lan9118_mac_changed(s); > } > @@ -937,7 +939,7 @@ static uint32_t do_mac_read(lan9118_state *s, int reg) > > static void lan9118_eeprom_cmd(lan9118_state *s, int cmd, int addr) > { > - s->e2p_cmd = (s->e2p_cmd & 0x10) | (cmd << 28) | addr; > + s->e2p_cmd = (s->e2p_cmd & E2P_CMD_MAC_ADDR_LOADED) | (cmd << 28) | addr; > switch (cmd) { > case 0: > s->e2p_data = s->eeprom[addr];