| To: | Stephen Hemminger <shemminger@xxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH] pktgen handle netdev device getting full. |
| From: | "David S. Miller" <davem@xxxxxxxxxxxxx> |
| Date: | Thu, 16 Sep 2004 16:20:30 -0700 |
| Cc: | greearb@xxxxxxxxxxxxxxx, Robert.Olsson@xxxxxxxxxxx, davem@xxxxxxxxxx, netdev@xxxxxxxxxxx, hadi@xxxxxxxx |
| In-reply-to: | <20040916162051.1b55ae5b@dell_ss3.pdx.osdl.net> |
| References: | <20040916144332.37f19fcb@dell_ss3.pdx.osdl.net> <414A1AD2.1090109@candelatech.com> <20040916155913.577b878b.davem@davemloft.net> <20040916162051.1b55ae5b@dell_ss3.pdx.osdl.net> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
On Thu, 16 Sep 2004 16:20:51 -0700 Stephen Hemminger <shemminger@xxxxxxxx> wrote: > Well the E1000 fill's it Tx ring and returns TX_BUSY, and on > SMP can also get TX_LOCKED now. If that happens, the message shows up. > Several other drivers don't set stopped when ring is almost full, but > rely on scheduler to do it for them. TX_LOCKED is special, and actually this points out that pktgen.c needs to change how it interprets hard_start_xmit() return values now that TX_LOCKED is possible too. I'm sorry, is that what your patch was doing? Then it was correct, please resend. |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH] pktgen handle netdev device getting full., Ben Greear |
|---|---|
| Next by Date: | Re: [PATCH] acenic - don't spin in hard_start_xmit when ring fills, David S. Miller |
| Previous by Thread: | Re: [PATCH] pktgen handle netdev device getting full., Stephen Hemminger |
| Next by Thread: | Re: [PATCH] pktgen handle netdev device getting full., Ben Greear |
| Indexes: | [Date] [Thread] [Top] [All Lists] |