Date: Fri, 12 Mar 2010 20:29:12 +0100 From: Attilio Rao <attilio@freebsd.org> To: Alexander Motin <mav@freebsd.org> Cc: freebsd-scsi@freebsd.org, "Justin T. Gibbs" <gibbs@freebsd.org>, mj@feral.com Subject: Re: How is supposed to be protected the units list? Message-ID: <3bbf2fe11003121129w69b6e48xfc3982bd9016bcce@mail.gmail.com> In-Reply-To: <4B9A91DA.7030107@FreeBSD.org> References: <3bbf2fe11002281655i61a5f0a0if3f381ad0c4a1ef8@mail.gmail.com> <3bbf2fe11003031532u2207eb55h19c3a045215a7d84@mail.gmail.com> <4B8EF336.80107@feral.com> <3bbf2fe11003031547kd5f7314t3d83b2bde06c1c2f@mail.gmail.com> <4B8EF990.5030407@feral.com> <3bbf2fe11003031607wa3727b5ke89bc2a909d4d6a6@mail.gmail.com> <4B901419.8060800@feral.com> <3bbf2fe11003041737p30690522ya81e1b8f4bd6bbf9@mail.gmail.com> <3bbf2fe11003120601y3c403a1ct50f9fc6c1f0903bf@mail.gmail.com> <4B9A91DA.7030107@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
2010/3/12 Alexander Motin <mav@freebsd.org>: > Attilio Rao wrote: >> 2010/3/5 Attilio Rao <attilio@freebsd.org>: >>> 2010/3/4 Matthew Jacob <mj@feral.com>: >>>> The referred to patch at least got me out of panic case :-).. >>>> >>>> http://people.freebsd.org/~mjacob/scsi_da.c.patch >>> Yes, honestly the main intent of this patch is to offer a stable >>> ground for correct handling of periph. When looking about refcounting >>> them correctly, the main problem is that there was no initial >>> condition assuring safety, and the initial patch should address this, >>> but I'm sure there are places where periph refcount is not handled >>> correctly and this may be one. >> >> So, as long as it seems nobody had a strong argument against this >> patch, what do you think about me committing it? >> We can further refine later if we think it is the case. >> >> Also, I think that Matt's patch should be committed just after this >> one (and possibly we should investigate a similar add-on for the ata >> counterpart too?). > > I have already told my opinion, that second lock may be not needed. I > would like to think a bit more about both patches after getting back > from the conference. Thanks, If you don't want to use the second lock, you have to use xpt_lock in all the other places anyways. I think it can get messy in particular if linked to another aspect: Another question, may be to integrate that locking paradigm directly with reference count of periph. It depends by if the periph can came from other parts as well. Attilio -- Peace can only be achieved by understanding - A. Einstein
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3bbf2fe11003121129w69b6e48xfc3982bd9016bcce>