From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49326) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bXlvQ-0005mB-7V for qemu-devel@nongnu.org; Thu, 11 Aug 2016 05:05:17 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bXlvM-0004wF-0l for qemu-devel@nongnu.org; Thu, 11 Aug 2016 05:05:15 -0400 Received: from mail-wm0-x243.google.com ([2a00:1450:400c:c09::243]:33545) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bXlvL-0004w9-MM for qemu-devel@nongnu.org; Thu, 11 Aug 2016 05:05:11 -0400 Received: by mail-wm0-x243.google.com with SMTP id o80so129373wme.0 for ; Thu, 11 Aug 2016 02:05:11 -0700 (PDT) Date: Thu, 11 Aug 2016 10:05:09 +0100 From: Stefan Hajnoczi Message-ID: <20160811090509.GD3932@stefanha-x1.localdomain> References: <1468340586-19304-1-git-send-email-pl@kamp.de> <1468340586-19304-2-git-send-email-pl@kamp.de> <20160808103750.GA30105@stefanha-x1.localdomain> <57A8CFA6.5070503@kamp.de> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="SO98HVl1bnMOfKZd" Content-Disposition: inline In-Reply-To: <57A8CFA6.5070503@kamp.de> Subject: Re: [Qemu-devel] [PATCH V5 1/6] oslib-posix: add helpers for stack alloc and free List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Lieven Cc: qemu-devel@nongnu.org, kwolf@redhat.com, peter.maydell@linaro.org, mst@redhat.com, armbru@redhat.com, dgilbert@redhat.com, mreitz@redhat.com, pbonzini@redhat.com, rth@twiddle.net --SO98HVl1bnMOfKZd Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Mon, Aug 08, 2016 at 08:29:58PM +0200, Peter Lieven wrote: > Am 08.08.2016 um 12:37 schrieb Stefan Hajnoczi: > > On Tue, Jul 12, 2016 at 06:23:01PM +0200, Peter Lieven wrote: > >> the allocated stack will be adjusted to the minimum supported stack si= ze > >> by the OS and rounded up to be a multiple of the system pagesize. > >> Additionally an architecture dependent guard page is added to the stack > >> to catch stack overflows. > >> > >> Signed-off-by: Peter Lieven > >> --- > >> include/sysemu/os-posix.h | 23 +++++++++++++++++++++++ > >> util/oslib-posix.c | 46 ++++++++++++++++++++++++++++++++++++++= ++++++++ > >> 2 files changed, 69 insertions(+) > >> > >> diff --git a/include/sysemu/os-posix.h b/include/sysemu/os-posix.h > >> index 9c7dfdf..7630665 100644 > >> --- a/include/sysemu/os-posix.h > >> +++ b/include/sysemu/os-posix.h > >> @@ -60,4 +60,27 @@ int qemu_utimens(const char *path, const qemu_times= pec *times); > >> =20 > >> bool is_daemonized(void); > >> =20 > >> +/** > >> + * qemu_alloc_stack: > >> + * @sz: size of required stack in bytes > >> + * > >> + * Allocate memory that can be used as a stack, for instance for > >> + * coroutines. If the memory cannot be allocated, this function > >> + * will abort (like g_malloc()). > >> + * > >> + * The allocated stack must be freed with qemu_free_stack(). > >> + * > >> + * Returns: pointer to (the lowest address of) the stack memory. > >> + */ > >> +void *qemu_alloc_stack(size_t sz); > >> + > >> +/** > >> + * qemu_free_stack: > >> + * @stack: stack to free > >> + * @sz: size of stack in bytes > >> + * > >> + * Free a stack allocated via qemu_alloc_stack(). > >> + */ > >> +void qemu_free_stack(void *stack, size_t sz); > >> + > >> #endif > >> diff --git a/util/oslib-posix.c b/util/oslib-posix.c > >> index e2e1d4d..2303ca6 100644 > >> --- a/util/oslib-posix.c > >> +++ b/util/oslib-posix.c > >> @@ -497,3 +497,49 @@ pid_t qemu_fork(Error **errp) > >> } > >> return pid; > >> } > >> + > >> +static size_t adjust_stack_size(size_t sz) > >> +{ > >> +#ifdef _SC_THREAD_STACK_MIN > >> + /* avoid stacks smaller than _SC_THREAD_STACK_MIN */ > >> + sz =3D MAX(MAX(sysconf(_SC_THREAD_STACK_MIN), 0), sz); > >> +#endif > >> + /* adjust stack size to a multiple of the page size */ > >> + sz =3D ROUND_UP(sz, getpagesize()); > >> + return sz; > >> +} > >> + > >> +void *qemu_alloc_stack(size_t sz) > >> +{ > >> + void *ptr, *guardpage; > >> + size_t pagesz =3D getpagesize(); > >> + sz =3D adjust_stack_size(sz); > >> + > >> + ptr =3D mmap(NULL, sz, PROT_READ | PROT_WRITE, > > It's cleaner to count for the guard page separately and give the caller > > the sz bytes they expected: > > > > sz + pagesz >=20 > I don't mind to change that, but the glibc stack allocation routine > does it exactly this way. (see glibc/nptl/allocatestack.c). > As we already used other parts (minimal stack size, positioning > of the guard page), we mimicked this as well. Okay but please include a comment explaining the rationale. Stefan --SO98HVl1bnMOfKZd Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQEcBAEBAgAGBQJXrD/FAAoJEJykq7OBq3PIrGsH/1LDrGsbkszNKT4ssoxCBpid e9W11QsJCKNY4dzluSFbQplzLOlJWtzCb7lh11fMnDT0nTBZx9vsTVxdcBLwnAOi Nywmz3MDXSCe5Uar0qKRhFR4nIvaNnlQBxNke78ohGStKUME4rSh/i8Dh8KBigG5 2qtFFGnUXhDN2rbETGRVaNEmwKh1zTa+VbZopOfwRlNVfbqV3Sa4C/z8eA2gIFq8 FcdjD5IFCDVXwgDO6LO1s0q0AKeHuvXGupAxho0jWtdUXpk3hGaR+iu4ZhvMMEDv botRzozxgGRhHD4nxQtsme0BbZ/t7WkE1nKoMzieUdJESVMX1SrotDU5mInppLQ= =lUcj -----END PGP SIGNATURE----- --SO98HVl1bnMOfKZd--