From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:50241) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YYOAh-0003dh-K8 for qemu-devel@nongnu.org; Wed, 18 Mar 2015 20:18:48 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YYOAY-0006bY-Kf for qemu-devel@nongnu.org; Wed, 18 Mar 2015 20:18:47 -0400 Received: from relay-06.andrew.cmu.edu ([128.2.157.21]:48121 helo=relay.andrew.cmu.edu) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YYOAY-0006bN-G5 for qemu-devel@nongnu.org; Wed, 18 Mar 2015 20:18:38 -0400 From: "Gabriel L. Somlo" Date: Wed, 18 Mar 2015 20:18:26 -0400 Message-Id: <1426724311-9343-1-git-send-email-somlo@cmu.edu> Subject: [Qemu-devel] [PATCH v2 0/5] fw-cfg: documentation, cleanup, and cmdline blobs List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: matt.fleming@intel.com, rjones@redhat.com, jordan.l.justen@intel.com, gleb@cloudius-systems.com, mdroth@linux.vnet.ibm.com, gsomlo@gmail.com, kraxel@redhat.com, pbonzini@redhat.com, lersek@redhat.com, armbru@redhat.com Document and clean up fw_cfg; additionally, allow user-provided blobs to be inserted into fw_cfg via the qemu command line. Changes since v1: - reworked documentation (1/5) as per Laszlo's feedback (actual changes detailed below the commit blurb in the actual patch) - guest-side write in 2/5 is now a no-op (as opposed to completely removing the write_op itself); also removed trace event. - reworded commit blurbs for 3/5 and 4/5; the latter now also removes a no longer needed trace event. - streamlined 5/5 (inserting fw_cfg blobs via command line) Specific changes and potentially remaining issues described at length below the commit blurb in the actual patch - dropped 6th patch (guest-side qga hack to retrieve fw_cfg files by name), as it is merely an RFC at this time, and we can continue talking about it in its original thread on the mailing list. I sent out a couple of emails about patch #5 earlier this afternoon, but please ignore that as I'm bringing up the same issues below the commit blurb, only hopefully in a way that's a bit more eloquent, or at least more coherent :) Thanks for all the feedback, and please let me know what you all think. Gabriel >Summary of patches in v1 of the series: > > 1/6: resubmit Jordan's documentation patch from back in 2011 > > 2/6: remove support for guest-side writes to fw_cfg data port > (this also updates the documentation applied earlier in 1/6) > > 3/6: add assertion to prevent us from inadvertently introducing memory leaks > when adding data blobs to fw_cfg > > 4/6: adding fw_cfg blobs with the same file name multiple times is a > memory-leaking erroneous thing to do: make qemu properly quit with > an error instead of just generating a trace event. > > 5/6: allow users to explicitly insert an arbitrary file as a fw_cfg blob > via the qemu command line > > 6/6: guest-side retrieval of named blob from the fw_cfg device > (please do NOT apply 6/6, as I'm just soliciting feedback at this time) Gabriel L. Somlo (5): fw_cfg: add documentation file (docs/specs/fw_cfg.txt) fw_cfg: remove support for guest-side data writes fw_cfg: prevent selector key conflict fw_cfg: prohibit insertion of duplicate fw_cfg file names fw_cfg: insert fw_cfg file blobs via qemu cmdline docs/specs/fw_cfg.txt | 192 ++++++++++++++++++++++++++++++++++++++++++++++ hw/nvram/fw_cfg.c | 85 +++++++++++--------- include/hw/nvram/fw_cfg.h | 6 +- qemu-options.hx | 11 +++ trace-events | 2 - vl.c | 27 +++++++ 6 files changed, 282 insertions(+), 41 deletions(-) create mode 100644 docs/specs/fw_cfg.txt -- 2.1.0