| Message ID | 1424982854.4444.73.camel@xylophone.i.decadent.org.uk |
|---|---|
| State | Accepted, archived |
| Delegated to: | David Miller |
| Headers | show |
Hello. On 2/26/2015 11:34 PM, Ben Hutchings wrote: > Currently we may silently read/write a register at offset 0. Change > this to WARN and then ignore the write or read-back all-ones. I think reading non-existing registers would yield all 0s, not all 1s. That's not x86. :-) > Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk> Other than that, seems a great idea. Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> WBR, Sergei -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Replying to a patchwork mbox, as I noticed this is in net-next. On Thu, 26 Feb 2015, Ben Hutchings wrote: > Currently we may silently read/write a register at offset 0. Change > this to WARN and then ignore the write or read-back all-ones. > > Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk> > Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> While this may be a good idea for debugging... > --- a/drivers/net/ethernet/renesas/sh_eth.h > +++ b/drivers/net/ethernet/renesas/sh_eth.h > @@ -543,19 +543,29 @@ static inline void sh_eth_soft_swap(char *src, int len) > #endif > } > > +#define SH_ETH_OFFSET_INVALID ((u16) ~0) > + > static inline void sh_eth_write(struct net_device *ndev, u32 data, > int enum_index) > { > struct sh_eth_private *mdp = netdev_priv(ndev); > + u16 offset = mdp->reg_offset[enum_index]; > + > + if (WARN_ON(offset == SH_ETH_OFFSET_INVALID)) > + return; ... adding WARN_ON() to static inline functions increases code size a lot: $ size drivers/net/ethernet/renesas/sh_eth.o{.orig,} text data bss dec hex filename 23352 1136 0 24488 5fa8 drivers/net/ethernet/renesas/sh_eth.o.orig 27225 1136 0 28361 6ec9 drivers/net/ethernet/renesas/sh_eth.o $ Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, 2015-03-05 at 10:02 +0100, Geert Uytterhoeven wrote: > Replying to a patchwork mbox, as I noticed this is in net-next. > > On Thu, 26 Feb 2015, Ben Hutchings wrote: > > Currently we may silently read/write a register at offset 0. Change > > this to WARN and then ignore the write or read-back all-ones. > > > > Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk> > > Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> > > While this may be a good idea for debugging... > > > --- a/drivers/net/ethernet/renesas/sh_eth.h > > +++ b/drivers/net/ethernet/renesas/sh_eth.h > > @@ -543,19 +543,29 @@ static inline void sh_eth_soft_swap(char *src, int len) > > #endif > > } > > > > +#define SH_ETH_OFFSET_INVALID ((u16) ~0) > > + > > static inline void sh_eth_write(struct net_device *ndev, u32 data, > > int enum_index) > > { > > struct sh_eth_private *mdp = netdev_priv(ndev); > > + u16 offset = mdp->reg_offset[enum_index]; > > + > > + if (WARN_ON(offset == SH_ETH_OFFSET_INVALID)) > > + return; > > ... adding WARN_ON() to static inline functions increases code size a lot: > > $ size drivers/net/ethernet/renesas/sh_eth.o{.orig,} > text data bss dec hex filename > 23352 1136 0 24488 5fa8 drivers/net/ethernet/renesas/sh_eth.o.orig > 27225 1136 0 28361 6ec9 drivers/net/ethernet/renesas/sh_eth.o > $ Time to un-inline it, maybe? Ben. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 05/03/15 13:18, Ben Hutchings wrote: > On Thu, 2015-03-05 at 10:02 +0100, Geert Uytterhoeven wrote: >> Replying to a patchwork mbox, as I noticed this is in net-next. >> >> On Thu, 26 Feb 2015, Ben Hutchings wrote: >>> Currently we may silently read/write a register at offset 0. Change >>> this to WARN and then ignore the write or read-back all-ones. >>> >>> Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk> >>> Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> >> >> While this may be a good idea for debugging... >> >>> --- a/drivers/net/ethernet/renesas/sh_eth.h >>> +++ b/drivers/net/ethernet/renesas/sh_eth.h >>> @@ -543,19 +543,29 @@ static inline void sh_eth_soft_swap(char *src, int len) >>> #endif >>> } >>> >>> +#define SH_ETH_OFFSET_INVALID ((u16) ~0) >>> + >>> static inline void sh_eth_write(struct net_device *ndev, u32 data, >>> int enum_index) >>> { >>> struct sh_eth_private *mdp = netdev_priv(ndev); does de-referencing this each time make a difference? Looks like it would have been easier to pass an "struct sh_eth_private" instead of the "struct net_device *ndev" >>> + u16 offset = mdp->reg_offset[enum_index]; >>> + >>> + if (WARN_ON(offset == SH_ETH_OFFSET_INVALID)) >>> + return; You could cange the mdp->reg_offset to an mdp->reg_pointer and make any invalid registers a NULL. This would at-least make it fail on an invalid access. >> ... adding WARN_ON() to static inline functions increases code size a lot: >> >> $ size drivers/net/ethernet/renesas/sh_eth.o{.orig,} >> text data bss dec hex filename >> 23352 1136 0 24488 5fa8 drivers/net/ethernet/renesas/sh_eth.o.orig >> 27225 1136 0 28361 6ec9 drivers/net/ethernet/renesas/sh_eth.o >> $ > > Time to un-inline it, maybe? > > Ben. > > > _______________________________________________ > linux-kernel mailing list > linux-kernel@lists.codethink.co.uk > https://ducie-dc1.codethink.co.uk/cgi-bin/mailman/listinfo/linux-kernel >
Hello. On 03/09/2015 11:50 AM, Ben Dooks wrote: >>>> Currently we may silently read/write a register at offset 0. Change >>>> this to WARN and then ignore the write or read-back all-ones. >>>> Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk> >>>> Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> >>> While this may be a good idea for debugging... >>>> --- a/drivers/net/ethernet/renesas/sh_eth.h >>>> +++ b/drivers/net/ethernet/renesas/sh_eth.h >>>> @@ -543,19 +543,29 @@ static inline void sh_eth_soft_swap(char *src, int len) >>>> #endif >>>> } >>>> >>>> +#define SH_ETH_OFFSET_INVALID ((u16) ~0) >>>> + >>>> static inline void sh_eth_write(struct net_device *ndev, u32 data, >>>> int enum_index) >>>> { >>>> struct sh_eth_private *mdp = netdev_priv(ndev); > does de-referencing this each time make a difference? Looks like It's not a dereference, it's just a pointer addition. > it would have been easier to pass an "struct sh_eth_private" instead > of the "struct net_device *ndev" Hm, maybe... >>>> + u16 offset = mdp->reg_offset[enum_index]; >>>> + >>>> + if (WARN_ON(offset == SH_ETH_OFFSET_INVALID)) >>>> + return; > You could cange the mdp->reg_offset to an mdp->reg_pointer and make > any invalid registers a NULL. This would at-least make it fail on an > invalid access. Interesting idea... WBR, Sergei -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/net/ethernet/renesas/sh_eth.c b/drivers/net/ethernet/renesas/sh_eth.c index 4169f93b5c3c..e21f8ce609cd 100644 --- a/drivers/net/ethernet/renesas/sh_eth.c +++ b/drivers/net/ethernet/renesas/sh_eth.c @@ -52,7 +52,12 @@ NETIF_MSG_RX_ERR| \ NETIF_MSG_TX_ERR) +#define SH_ETH_OFFSET_DEFAULTS \ + [0 ... SH_ETH_MAX_REGISTER_OFFSET - 1] = SH_ETH_OFFSET_INVALID + static const u16 sh_eth_offset_gigabit[SH_ETH_MAX_REGISTER_OFFSET] = { + SH_ETH_OFFSET_DEFAULTS, + [EDSR] = 0x0000, [EDMR] = 0x0400, [EDTRR] = 0x0408, @@ -151,6 +156,8 @@ static const u16 sh_eth_offset_gigabit[SH_ETH_MAX_REGISTER_OFFSET] = { }; static const u16 sh_eth_offset_fast_rz[SH_ETH_MAX_REGISTER_OFFSET] = { + SH_ETH_OFFSET_DEFAULTS, + [EDSR] = 0x0000, [EDMR] = 0x0400, [EDTRR] = 0x0408, @@ -210,6 +217,8 @@ static const u16 sh_eth_offset_fast_rz[SH_ETH_MAX_REGISTER_OFFSET] = { }; static const u16 sh_eth_offset_fast_rcar[SH_ETH_MAX_REGISTER_OFFSET] = { + SH_ETH_OFFSET_DEFAULTS, + [ECMR] = 0x0300, [RFLR] = 0x0308, [ECSR] = 0x0310, @@ -256,6 +265,8 @@ static const u16 sh_eth_offset_fast_rcar[SH_ETH_MAX_REGISTER_OFFSET] = { }; static const u16 sh_eth_offset_fast_sh4[SH_ETH_MAX_REGISTER_OFFSET] = { + SH_ETH_OFFSET_DEFAULTS, + [ECMR] = 0x0100, [RFLR] = 0x0108, [ECSR] = 0x0110, @@ -308,6 +319,8 @@ static const u16 sh_eth_offset_fast_sh4[SH_ETH_MAX_REGISTER_OFFSET] = { }; static const u16 sh_eth_offset_fast_sh3_sh2[SH_ETH_MAX_REGISTER_OFFSET] = { + SH_ETH_OFFSET_DEFAULTS, + [EDMR] = 0x0000, [EDTRR] = 0x0004, [EDRRR] = 0x0008, @@ -1541,7 +1554,8 @@ static int sh_eth_rx(struct net_device *ndev, u32 intr_status, int *quota) /* If we don't need to check status, don't. -KDU */ if (!(sh_eth_read(ndev, EDRRR) & EDRRR_R)) { /* fix the values for the next receiving if RDE is set */ - if (intr_status & EESR_RDE && mdp->reg_offset[RDFAR] != 0) { + if (intr_status & EESR_RDE && + mdp->reg_offset[RDFAR] != SH_ETH_OFFSET_INVALID) { u32 count = (sh_eth_read(ndev, RDFAR) - sh_eth_read(ndev, RDLAR)) >> 4; diff --git a/drivers/net/ethernet/renesas/sh_eth.h b/drivers/net/ethernet/renesas/sh_eth.h index 259d03f353e1..33a360c4fd10 100644 --- a/drivers/net/ethernet/renesas/sh_eth.h +++ b/drivers/net/ethernet/renesas/sh_eth.h @@ -543,19 +543,29 @@ static inline void sh_eth_soft_swap(char *src, int len) #endif } +#define SH_ETH_OFFSET_INVALID ((u16) ~0) + static inline void sh_eth_write(struct net_device *ndev, u32 data, int enum_index) { struct sh_eth_private *mdp = netdev_priv(ndev); + u16 offset = mdp->reg_offset[enum_index]; + + if (WARN_ON(offset == SH_ETH_OFFSET_INVALID)) + return; - iowrite32(data, mdp->addr + mdp->reg_offset[enum_index]); + iowrite32(data, mdp->addr + offset); } static inline u32 sh_eth_read(struct net_device *ndev, int enum_index) { struct sh_eth_private *mdp = netdev_priv(ndev); + u16 offset = mdp->reg_offset[enum_index]; + + if (WARN_ON(offset == SH_ETH_OFFSET_INVALID)) + return ~0U; - return ioread32(mdp->addr + mdp->reg_offset[enum_index]); + return ioread32(mdp->addr + offset); } static inline void *sh_eth_tsu_get_offset(struct sh_eth_private *mdp,
Currently we may silently read/write a register at offset 0. Change this to WARN and then ignore the write or read-back all-ones. Signed-off-by: Ben Hutchings <ben.hutchings@codethink.co.uk> --- drivers/net/ethernet/renesas/sh_eth.c | 16 +++++++++++++++- drivers/net/ethernet/renesas/sh_eth.h | 14 ++++++++++++-- 2 files changed, 27 insertions(+), 3 deletions(-)