linux-imx
Public
Source
committed 0892fac8719x
static int bm_rw(struct drbd_device *device, const unsigned int flags, unsigned lazy_writeout_upper_idx) __must_hold(local)/* drbd_bitmap.c This file is part of DRBD by Philipp Reisner and Lars Ellenberg. Copyright (C) 2004-2008, LINBIT Information Technologies GmbH. Copyright (C) 2004-2008, Philipp Reisner <philipp.reisner@linbit.com>. Copyright (C) 2004-2008, Lars Ellenberg <lars.ellenberg@linbit.com>. drbd is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. drbd is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with drbd; see the file COPYING. If not, write to the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. *//* OPAQUE outside this file! * interface defined in drbd_int.h * convention: * function name drbd_bm_... => used elsewhere, "public". * function name bm_... => internal to implementation, "private". *//* * LIMITATIONS: * We want to support >= peta byte of backend storage, while for now still using * a granularity of one bit per 4KiB of storage. * 1 << 50 bytes backend storage (1 PiB) * 1 << (50 - 12) bits needed * 38 --> we need u64 to index and count bits * 1 << (38 - 3) bitmap bytes needed * 35 --> we still need u64 to index and count bytes * (that's 32 GiB of bitmap for 1 PiB storage) * 1 << (35 - 2) 32bit longs needed * 33 --> we'd even need u64 to index and count 32bit long words. * 1 << (35 - 3) 64bit longs needed * 32 --> we could get away with a 32bit unsigned int to index and count * 64bit long words, but I rather stay with unsigned long for now. * We probably should neither count nor point to bytes or long words * directly, but either by bitnumber, or by page index and offset. * 1 << (35 - 12) * 22 --> we need that much 4KiB pages of bitmap. * 1 << (22 + 3) --> on a 64bit arch, * we need 32 MiB to store the array of page pointers. * * Because I'm lazy, and because the resulting patch was too large, too ugly * and still incomplete, on 32bit we still "only" support 16 TiB (minus some), * (1 << 32) bits * 4k storage. * * bitmap storage and IO: * Bitmap is stored little endian on disk, and is kept little endian in * core memory. Currently we still hold the full bitmap in core as long * as we are "attached" to a local disk, which at 32 GiB for 1PiB storage * seems excessive. * * We plan to reduce the amount of in-core bitmap pages by paging them in * and out against their on-disk location as necessary, but need to make * sure we don't cause too much meta data IO, and must not deadlock in * tight memory situations. This needs some more work. *//* * NOTE * Access to the *bm_pages is protected by bm_lock. * It is safe to read the other members within the lock. * * drbd_bm_set_bits is called from bio_endio callbacks, * We may be called with irq already disabled, * so we need spin_lock_irqsave().