From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1O3pb4-0006Vw-Fk for qemu-devel@nongnu.org; Mon, 19 Apr 2010 07:57:02 -0400 Received: from [140.186.70.92] (port=43060 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1O3pau-0006Qf-QT for qemu-devel@nongnu.org; Mon, 19 Apr 2010 07:57:01 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.69) (envelope-from ) id 1O3par-0003u0-El for qemu-devel@nongnu.org; Mon, 19 Apr 2010 07:56:52 -0400 Received: from mx1.redhat.com ([209.132.183.28]:30556) by eggs.gnu.org with esmtp (Exim 4.69) (envelope-from ) id 1O3par-0003tp-6c for qemu-devel@nongnu.org; Mon, 19 Apr 2010 07:56:49 -0400 Message-ID: <4BCC44FD.8000008@redhat.com> Date: Mon, 19 Apr 2010 14:56:45 +0300 From: Avi Kivity MIME-Version: 1.0 Subject: Re: [Qemu-devel] Re: [PATCH v3 2/6] Introduce bit-based phys_ram_dirty for VGA, CODE, MIGRATION and MASTER. References: <1271670198-12793-1-git-send-email-tamura.yoshiaki@lab.ntt.co.jp> <1271670198-12793-3-git-send-email-tamura.yoshiaki@lab.ntt.co.jp> <4BCC2DBA.4000602@redhat.com> <4BCC3226.20305@lab.ntt.co.jp> <4BCC40C6.1040203@redhat.com> <4BCC43FA.3090704@lab.ntt.co.jp> In-Reply-To: <4BCC43FA.3090704@lab.ntt.co.jp> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Yoshiaki Tamura Cc: aliguori@us.ibm.com, mtosatti@redhat.com, qemu-devel@nongnu.org, ohmura.kei@lab.ntt.co.jp On 04/19/2010 02:52 PM, Yoshiaki Tamura wrote: > Avi Kivity wrote: >> On 04/19/2010 02:31 PM, Yoshiaki Tamura wrote: >>> >>>>>> typedef struct RAMBlock { >>>>>> @@ -2825,10 +2825,16 @@ ram_addr_t qemu_ram_alloc(ram_addr_t size) >>>>>> new_block->next = ram_blocks; >>>>>> ram_blocks = new_block; >>>>>> >>>>>> - phys_ram_dirty = qemu_realloc(phys_ram_dirty, >>>>>> - (last_ram_offset + size)>> TARGET_PAGE_BITS); >>>>>> - memset(phys_ram_dirty + (last_ram_offset>> TARGET_PAGE_BITS), >>>>>> - 0xff, size>> TARGET_PAGE_BITS); >>>>>> + if (BITMAP_SIZE(last_ram_offset + size) != >>>>>> BITMAP_SIZE(last_ram_offset)) { >>>>> This check is unneeded - the code will work fine even if the bitmap >>>>> size >>>>> doesn't change. >>>> OK. I'll remove it. >>> I have a problem here. >>> If I remove this check, glibc reports an error as below. >>> >>> *** glibc detected *** /usr/local/qemu/bin/qemu-system-x86_64: >>> realloc(): invalid pointer: 0x0000000001f0e450 *** >>> ======= Backtrace: ========= >>> /lib64/libc.so.6[0x369fa75a96] >>> /lib64/libc.so.6(realloc+0x2a1)[0x369fa7b881] >>> /usr/local/qemu/bin/qemu-system-x86_64[0x437d93] >>> /usr/local/qemu/bin/qemu-system-x86_64[0x4f03f6] >>> /usr/local/qemu/bin/qemu-system-x86_64[0x5b052c] >>> /usr/local/qemu/bin/qemu-system-x86_64[0x5b0d8b] >>> /usr/local/qemu/bin/qemu-system-x86_64[0x41ec2b] >>> /lib64/libc.so.6(__libc_start_main+0xfd)[0x369fa1ea2d] >>> /usr/local/qemu/bin/qemu-system-x86_64[0x406479] >>> ======= Memory map: ======== >>> >>> I reminded that I put this check to avoid reallocating same size to >>> the bitmap. >>> qemu goes this routine at start up, and extends last_ram_offset at >>> small numbers. >>> The error above is reported at the extension phase. >>> >> >> This probably means that an old bitmap pointer leaked somewhere, and we >> realloc() it after free? Or perhaps a glibc bug. > > Original qemu doesn't have a code the frees phys_ram_dirty, and I > didn't either. > Hmmm. I meant, after we realloc() something we keep using the old pointer. realloc() is equivalent to free(), after all. It might also be memory corruption - bits set outside the bitmap and hitting glibc malloc metadata. -- error compiling committee.c: too many arguments to function