From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:57048) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZlqPB-0002nX-Ux for qemu-devel@nongnu.org; Mon, 12 Oct 2015 23:37:38 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZlqP8-0008GZ-NO for qemu-devel@nongnu.org; Mon, 12 Oct 2015 23:37:37 -0400 Received: from mga09.intel.com ([134.134.136.24]:22949) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZlqP8-0008GT-GZ for qemu-devel@nongnu.org; Mon, 12 Oct 2015 23:37:34 -0400 References: <1444535584-18220-1-git-send-email-guangrong.xiao@linux.intel.com> <1444535584-18220-9-git-send-email-guangrong.xiao@linux.intel.com> <20151012100842.GA1681@redhat.com> From: Xiao Guangrong Message-ID: <561C7B0A.9050303@linux.intel.com> Date: Tue, 13 Oct 2015 11:31:22 +0800 MIME-Version: 1.0 In-Reply-To: <20151012100842.GA1681@redhat.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit Subject: Re: [Qemu-devel] [PATCH v3 08/32] exec: allow memory to be allocated from any kind of path List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Michael S. Tsirkin" Cc: ehabkost@redhat.com, kvm@vger.kernel.org, gleb@kernel.org, mtosatti@redhat.com, qemu-devel@nongnu.org, stefanha@redhat.com, imammedo@redhat.com, pbonzini@redhat.com, dan.j.williams@intel.com, rth@twiddle.net On 10/12/2015 06:08 PM, Michael S. Tsirkin wrote: > On Sun, Oct 11, 2015 at 11:52:40AM +0800, Xiao Guangrong wrote: >> Currently file_ram_alloc() is designed for hugetlbfs, however, the memory >> of nvdimm can come from either raw pmem device eg, /dev/pmem, or the file >> locates at DAX enabled filesystem >> >> So this patch let it work on any kind of path >> >> Signed-off-by: Xiao Guangrong > > This conflicts with map alloc rework. > Please rebase this on top of my tree. > Okay, thanks for your reminder. I did it based on upstream QEMU tree, will do it on pci branch on your tree instead. > >> --- >> exec.c | 55 ++++++++++++++----------------------------------------- >> 1 file changed, 14 insertions(+), 41 deletions(-) >> >> diff --git a/exec.c b/exec.c >> index 7d90a52..70cb0ef 100644 >> --- a/exec.c >> +++ b/exec.c >> @@ -1154,32 +1154,6 @@ void qemu_mutex_unlock_ramlist(void) >> } >> >> #ifdef __linux__ >> - >> -#include >> - >> -#define HUGETLBFS_MAGIC 0x958458f6 >> - >> -static long gethugepagesize(const char *path, Error **errp) >> -{ >> - struct statfs fs; >> - int ret; >> - >> - do { >> - ret = statfs(path, &fs); >> - } while (ret != 0 && errno == EINTR); >> - >> - if (ret != 0) { >> - error_setg_errno(errp, errno, "failed to get page size of file %s", >> - path); >> - return 0; >> - } >> - >> - if (fs.f_type != HUGETLBFS_MAGIC) >> - fprintf(stderr, "Warning: path not on HugeTLBFS: %s\n", path); >> - >> - return fs.f_bsize; > > What this *actually* is trying to warn against is that > mapping a regular file (as opposed to hugetlbfs) > means transparent huge pages don't work. > > So I don't think we should drop this warning completely. > Either let's add the nvdimm magic, or simply check the > page size. Check the page size sounds good, will check: if (pagesize != getpagesize()) { ...print something... } I agree with you that showing the info is needed, however, 'Warning' might scare some users, how about drop this word or just show　“Memory is not allocated from HugeTlbfs”?