| Message ID | 20140701190347.GU1688@ERROL.INI.CMU.EDU |
|---|---|
| State | New |
| Headers | show |
On Tue, Jul 01, 2014 at 03:03:48PM -0400, Gabriel L. Somlo wrote: > Make phyreg_writeops responsible for actually writing their > respective phy registers. The only current instance of > phyreg_writeops is set_phy_ctrl(), which we modify to actually > write the register, while also correctly handling reserved and > self-clearing bits. > > have_autoneg() does not need to check for MII_CR_RESTART_AUTO_NEG, > since the only time the flag comes into play is during set_phy_ctrl(), > and never actually gets written to the phy control register. We also > move it in front of set_phy_ctrl(), to avoid a forward declaration. > > Signed-off-by: Gabriel Somlo <somlo@cmu.edu> > --- Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Applied, thanks! > Michael, Stefan: > > Here's a new version of this patch with slightly better (IMHO) formatting. > I think the merits are 1. handling bits which are reserved or self > clearing; and 2. eliminating the surprise of a "write op" that does not > actually write to its register :) > > Regarding the "sibling" patch originally accompanying this one (which > increased the timer delay for autonegotiation), I'll take a while > longer to figure out what's really going on, so I'll send out a separate > patch if/when that happens. > > Please apply (or review/critique) this one independently. > > Thanks, > Gabriel > > hw/net/e1000.c | 31 +++++++++++++++++-------------- > 1 file changed, 17 insertions(+), 14 deletions(-) > > diff --git a/hw/net/e1000.c b/hw/net/e1000.c > index 0fc29a0..04c0f91 100644 > --- a/hw/net/e1000.c > +++ b/hw/net/e1000.c > @@ -186,21 +186,31 @@ e1000_link_up(E1000State *s) > s->phy_reg[PHY_STATUS] |= MII_SR_LINK_STATUS; > } > > +static bool > +have_autoneg(E1000State *s) > +{ > + return (s->compat_flags & E1000_FLAG_AUTONEG) && > + (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN); > +} > + > static void > set_phy_ctrl(E1000State *s, int index, uint16_t val) > { > + /* bits 0-5 reserved; MII_CR_[RESTART_AUTO_NEG,RESET] are self clearing */ > + s->phy_reg[PHY_CTRL] = val & ~(0x3f | > + MII_CR_RESET | > + MII_CR_RESTART_AUTO_NEG); > + > /* > * QEMU 1.3 does not support link auto-negotiation emulation, so if we > * migrate during auto negotiation, after migration the link will be > * down. > */ > - if (!(s->compat_flags & E1000_FLAG_AUTONEG)) { > - return; > - } > - if ((val & MII_CR_AUTO_NEG_EN) && (val & MII_CR_RESTART_AUTO_NEG)) { > + if (have_autoneg(s) && (val & MII_CR_RESTART_AUTO_NEG)) { > e1000_link_down(s); > DBGOUT(PHY, "Start link auto negotiation\n"); > - timer_mod(s->autoneg_timer, qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500); > + timer_mod(s->autoneg_timer, > + qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500); > } > } > > @@ -446,8 +456,9 @@ set_mdic(E1000State *s, int index, uint32_t val) > } else { > if (addr < NPHYWRITEOPS && phyreg_writeops[addr]) { > phyreg_writeops[addr](s, index, data); > + } else { > + s->phy_reg[addr] = data; > } > - s->phy_reg[addr] = data; > } > } > s->mac_reg[MDIC] = val | E1000_MDIC_READY; > @@ -848,14 +859,6 @@ receive_filter(E1000State *s, const uint8_t *buf, int size) > return 0; > } > > -static bool > -have_autoneg(E1000State *s) > -{ > - return (s->compat_flags & E1000_FLAG_AUTONEG) && > - (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN) && > - (s->phy_reg[PHY_CTRL] & MII_CR_RESTART_AUTO_NEG); > -} > - > static void > e1000_set_link_status(NetClientState *nc) > { > -- > 1.9.3
On Wed, Jul 02, 2014 at 12:12:30PM +0300, Michael S. Tsirkin wrote: > On Tue, Jul 01, 2014 at 03:03:48PM -0400, Gabriel L. Somlo wrote: > > Make phyreg_writeops responsible for actually writing their > > respective phy registers. The only current instance of > > phyreg_writeops is set_phy_ctrl(), which we modify to actually > > write the register, while also correctly handling reserved and > > self-clearing bits. > > > > have_autoneg() does not need to check for MII_CR_RESTART_AUTO_NEG, > > since the only time the flag comes into play is during set_phy_ctrl(), > > and never actually gets written to the phy control register. We also > > move it in front of set_phy_ctrl(), to avoid a forward declaration. > > > > Signed-off-by: Gabriel Somlo <somlo@cmu.edu> > > --- > > > Reviewed-by: Michael S. Tsirkin <mst@redhat.com> > > Applied, thanks! And just to clarify (I'd like to add this to the commit log) we are fixing a minor bug here: MII_CR_RESTART_AUTO_NEG and MII_CR_RESET aren't self-clearing as they should be, and low 5 bits aren't RO as they should be. > > Michael, Stefan: > > > > Here's a new version of this patch with slightly better (IMHO) formatting. > > I think the merits are 1. handling bits which are reserved or self > > clearing; and 2. eliminating the surprise of a "write op" that does not > > actually write to its register :) > > > > Regarding the "sibling" patch originally accompanying this one (which > > increased the timer delay for autonegotiation), I'll take a while > > longer to figure out what's really going on, so I'll send out a separate > > patch if/when that happens. > > > > Please apply (or review/critique) this one independently. > > > > Thanks, > > Gabriel > > > > hw/net/e1000.c | 31 +++++++++++++++++-------------- > > 1 file changed, 17 insertions(+), 14 deletions(-) > > > > diff --git a/hw/net/e1000.c b/hw/net/e1000.c > > index 0fc29a0..04c0f91 100644 > > --- a/hw/net/e1000.c > > +++ b/hw/net/e1000.c > > @@ -186,21 +186,31 @@ e1000_link_up(E1000State *s) > > s->phy_reg[PHY_STATUS] |= MII_SR_LINK_STATUS; > > } > > > > +static bool > > +have_autoneg(E1000State *s) > > +{ > > + return (s->compat_flags & E1000_FLAG_AUTONEG) && > > + (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN); > > +} > > + > > static void > > set_phy_ctrl(E1000State *s, int index, uint16_t val) > > { > > + /* bits 0-5 reserved; MII_CR_[RESTART_AUTO_NEG,RESET] are self clearing */ > > + s->phy_reg[PHY_CTRL] = val & ~(0x3f | > > + MII_CR_RESET | > > + MII_CR_RESTART_AUTO_NEG); > > + > > /* > > * QEMU 1.3 does not support link auto-negotiation emulation, so if we > > * migrate during auto negotiation, after migration the link will be > > * down. > > */ > > - if (!(s->compat_flags & E1000_FLAG_AUTONEG)) { > > - return; > > - } > > - if ((val & MII_CR_AUTO_NEG_EN) && (val & MII_CR_RESTART_AUTO_NEG)) { > > + if (have_autoneg(s) && (val & MII_CR_RESTART_AUTO_NEG)) { > > e1000_link_down(s); > > DBGOUT(PHY, "Start link auto negotiation\n"); > > - timer_mod(s->autoneg_timer, qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500); > > + timer_mod(s->autoneg_timer, > > + qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500); > > } > > } > > > > @@ -446,8 +456,9 @@ set_mdic(E1000State *s, int index, uint32_t val) > > } else { > > if (addr < NPHYWRITEOPS && phyreg_writeops[addr]) { > > phyreg_writeops[addr](s, index, data); > > + } else { > > + s->phy_reg[addr] = data; > > } > > - s->phy_reg[addr] = data; > > } > > } > > s->mac_reg[MDIC] = val | E1000_MDIC_READY; > > @@ -848,14 +859,6 @@ receive_filter(E1000State *s, const uint8_t *buf, int size) > > return 0; > > } > > > > -static bool > > -have_autoneg(E1000State *s) > > -{ > > - return (s->compat_flags & E1000_FLAG_AUTONEG) && > > - (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN) && > > - (s->phy_reg[PHY_CTRL] & MII_CR_RESTART_AUTO_NEG); > > -} > > - > > static void > > e1000_set_link_status(NetClientState *nc) > > { > > -- > > 1.9.3
diff --git a/hw/net/e1000.c b/hw/net/e1000.c index 0fc29a0..04c0f91 100644 --- a/hw/net/e1000.c +++ b/hw/net/e1000.c @@ -186,21 +186,31 @@ e1000_link_up(E1000State *s) s->phy_reg[PHY_STATUS] |= MII_SR_LINK_STATUS; } +static bool +have_autoneg(E1000State *s) +{ + return (s->compat_flags & E1000_FLAG_AUTONEG) && + (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN); +} + static void set_phy_ctrl(E1000State *s, int index, uint16_t val) { + /* bits 0-5 reserved; MII_CR_[RESTART_AUTO_NEG,RESET] are self clearing */ + s->phy_reg[PHY_CTRL] = val & ~(0x3f | + MII_CR_RESET | + MII_CR_RESTART_AUTO_NEG); + /* * QEMU 1.3 does not support link auto-negotiation emulation, so if we * migrate during auto negotiation, after migration the link will be * down. */ - if (!(s->compat_flags & E1000_FLAG_AUTONEG)) { - return; - } - if ((val & MII_CR_AUTO_NEG_EN) && (val & MII_CR_RESTART_AUTO_NEG)) { + if (have_autoneg(s) && (val & MII_CR_RESTART_AUTO_NEG)) { e1000_link_down(s); DBGOUT(PHY, "Start link auto negotiation\n"); - timer_mod(s->autoneg_timer, qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500); + timer_mod(s->autoneg_timer, + qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500); } } @@ -446,8 +456,9 @@ set_mdic(E1000State *s, int index, uint32_t val) } else { if (addr < NPHYWRITEOPS && phyreg_writeops[addr]) { phyreg_writeops[addr](s, index, data); + } else { + s->phy_reg[addr] = data; } - s->phy_reg[addr] = data; } } s->mac_reg[MDIC] = val | E1000_MDIC_READY; @@ -848,14 +859,6 @@ receive_filter(E1000State *s, const uint8_t *buf, int size) return 0; } -static bool -have_autoneg(E1000State *s) -{ - return (s->compat_flags & E1000_FLAG_AUTONEG) && - (s->phy_reg[PHY_CTRL] & MII_CR_AUTO_NEG_EN) && - (s->phy_reg[PHY_CTRL] & MII_CR_RESTART_AUTO_NEG); -} - static void e1000_set_link_status(NetClientState *nc) {
Make phyreg_writeops responsible for actually writing their respective phy registers. The only current instance of phyreg_writeops is set_phy_ctrl(), which we modify to actually write the register, while also correctly handling reserved and self-clearing bits. have_autoneg() does not need to check for MII_CR_RESTART_AUTO_NEG, since the only time the flag comes into play is during set_phy_ctrl(), and never actually gets written to the phy control register. We also move it in front of set_phy_ctrl(), to avoid a forward declaration. Signed-off-by: Gabriel Somlo <somlo@cmu.edu> --- Michael, Stefan: Here's a new version of this patch with slightly better (IMHO) formatting. I think the merits are 1. handling bits which are reserved or self clearing; and 2. eliminating the surprise of a "write op" that does not actually write to its register :) Regarding the "sibling" patch originally accompanying this one (which increased the timer delay for autonegotiation), I'll take a while longer to figure out what's really going on, so I'll send out a separate patch if/when that happens. Please apply (or review/critique) this one independently. Thanks, Gabriel hw/net/e1000.c | 31 +++++++++++++++++-------------- 1 file changed, 17 insertions(+), 14 deletions(-)