Date: Wed, 30 Oct 2002 12:51:27 -0800 From: Luigi Rizzo <rizzo@icir.org> To: Ian Dowse <iedowse@maths.tcd.ie> Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/dev/fxp if_fxp.c Message-ID: <20021030125127.A25286@carp.icir.org> In-Reply-To: <200210302047.aa48208@salmon.maths.tcd.ie>; from iedowse@maths.tcd.ie on Wed, Oct 30, 2002 at 08:47:10PM %2B0000 References: <20021030122423.B24962@carp.icir.org> <200210302047.aa48208@salmon.maths.tcd.ie>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Oct 30, 2002 at 08:47:10PM +0000, Ian Dowse wrote: ... > Sorry, you're right there - the bit about it writing at the same > time as the hardware is incorrect. However if the 0x4000 bit gets > set (and there is no documentation that says it won't be), it is > possible for the RNR recovery code to be executed while the hardware > is still receiving packets: ok, this is another story and it might be something worth looking at. However i believe it cannot happen on the non-polling case so it still would not explain why Pawel's box was crashing, and furthermore, there are quite a few people using polling with the fxp on heavily loaded router boxes, just have a look at the zebra mailing lists, so i would say that if there is a problem in that code, it would have been reported.) cheers luigi > fxp_scb_wait(sc); > CSR_WRITE_4(sc, FXP_CSR_SCB_GENERAL, > vtophys(sc->rfa_headm->m_ext.ext_buf) + > RFA_ALIGNMENT_FUDGE); > fxp_scb_cmd(sc, FXP_SCB_COMMAND_RU_START); > > This is probably not good. > > Ian To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20021030125127.A25286>