From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:34239) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1eOQ95-0001d7-9m for qemu-devel@nongnu.org; Mon, 11 Dec 2017 10:37:32 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1eOQ94-0004nO-Dw for qemu-devel@nongnu.org; Mon, 11 Dec 2017 10:37:31 -0500 Received: from mail-ot0-x231.google.com ([2607:f8b0:4003:c0f::231]:45110) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1eOQ94-0004mm-7Y for qemu-devel@nongnu.org; Mon, 11 Dec 2017 10:37:30 -0500 Received: by mail-ot0-x231.google.com with SMTP id 103so15046284otj.12 for ; Mon, 11 Dec 2017 07:37:30 -0800 (PST) MIME-Version: 1.0 In-Reply-To: References: <1507824216-29058-1-git-send-email-peter.maydell@linaro.org> <1507824216-29058-9-git-send-email-peter.maydell@linaro.org> <20171208230920.GA11462@flamenco> From: Peter Maydell Date: Mon, 11 Dec 2017 15:37:09 +0000 Message-ID: Content-Type: text/plain; charset="UTF-8" Subject: Re: [Qemu-devel] [PULL 08/13] target/arm: Pull Thumb insn word loads up to top level List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Emilio G. Cota" Cc: QEMU Developers , Richard Henderson , qemu-arm On 10 December 2017 at 18:24, Peter Maydell wrote: > On 8 December 2017 at 23:09, Emilio G. Cota wrote: >> On Thu, Oct 12, 2017 at 17:03:31 +0100, Peter Maydell wrote: >>> Refactor the Thumb decode to do the loads of the instruction words at >>> the top level rather than only loading the second half of a 32-bit >>> Thumb insn in the middle of the decode. > >> This commit breaks the debian-arm boot test (see [1]), boot dies at: >> >>> random: systemd urandom read with 4 bits of entropy available >>> systemd[1]: Caught , core dump failed. >>> systemd[1]: Freezing execution. >> >> Sorry I noticed this so late (-rc4), for a while I thought my development >> was causing this and didn't pay much attention to it. However, just realised >> the problem is present on master. Bisect log below. > > It's missed the deadline for getting fixed in 2.11, but I'll have > a look at what's going on here. This is a really bad bug -- we accidentally stopped generating UNDEF for 32-bit Thumb2 instructions that should UNDEF (including things like "first use of FPU with FPU disabled", which is probably why systemd falls over). I wish we'd noticed this a week ago :-( Patch coming in a moment, anyway. thanks -- PMM