On Mon, 2013-02-04 at 16:36 +0100, Johannes Berg wrote:
> > + if ((sinfo->filled & STATION_INFO_RX_BYTES64) &&
> > + nla_put_u64(msg, NL80211_STA_INFO_RX_BYTES64,
> > + sinfo->rx_bytes))
> > + goto nla_put_failure;
> >[...]
>
> + if (sinfo->filled & STATION_INFO_RX_BYTES64) {
> + if (nla_put_u64(msg, NL80211_STA_INFO_RX_BYTES64,
> + sinfo->rx_bytes))
> + goto nla_put_failure;
> + if (sinfo->rx_bytes <= UINT_MAX &&
> + nla_put_u32(msg, NL80211_STA_INFO_RX_BYTES,
> + (u32)sinfo->rx_bytes))
> + goto nla_put_failure;
> + }
> + if (sinfo->filled & STATION_INFO_TX_BYTES64) {
> + if (nla_put_u64(msg, NL80211_STA_INFO_TX_BYTES64,
> + sinfo->tx_bytes))
> + goto nla_put_failure;
> + if (sinfo->tx_bytes <= UINT_MAX &&
> + nla_put_u32(msg, NL80211_STA_INFO_TX_BYTES,
> + (u32)sinfo->tx_bytes))
> + goto nla_put_failure;
> + }
Ok Vladimir convinced me this was stupid because it breaks the current
userspace API in a subtle way: right now, if it rolls over, we will get
wrapped values in userspace, afterwards with a 64-bit compatible driver
we won't get any 32-bit value ...
I'll change it to this though, so drivers don't get this choice:
- if ((sinfo->filled & STATION_INFO_RX_BYTES) &&
+ if ((sinfo->filled & (STATION_INFO_RX_BYTES |
+ STATION_INFO_RX_BYTES64)) &&
nla_put_u32(msg, NL80211_STA_INFO_RX_BYTES,
- sinfo->rx_bytes))
+ (u32)sinfo->rx_bytes))
goto nla_put_failure;
- if ((sinfo->filled & STATION_INFO_TX_BYTES) &&
+ if ((sinfo->filled & (STATION_INFO_TX_BYTES |
+ NL80211_STA_INFO_TX_BYTES64)) &&
nla_put_u32(msg, NL80211_STA_INFO_TX_BYTES,
+ (u32)sinfo->tx_bytes))
+ goto nla_put_failure;
+ if ((sinfo->filled & STATION_INFO_RX_BYTES64) &&
+ nla_put_u64(msg, NL80211_STA_INFO_RX_BYTES64,
+ sinfo->rx_bytes))
+ goto nla_put_failure;
+ if ((sinfo->filled & STATION_INFO_TX_BYTES64) &&
+ nla_put_u64(msg, NL80211_STA_INFO_TX_BYTES64,
sinfo->tx_bytes))
goto nla_put_failure;
IOW, always put the 32-bit value even if the driver said it had 64-bit,
but then also give the 64-bit value to userspace.
johannes
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
