linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Al Viro <viro@zeniv.linux.org.uk>,
	Guenter Roeck <linux@roeck-us.net>
Subject: [PATCH 5.0 64/89] pin iocb through aio.
Date: Tue, 30 Apr 2019 13:38:55 +0200	[thread overview]
Message-ID: <20190430113612.718349457@linuxfoundation.org> (raw)
In-Reply-To: <20190430113609.741196396@linuxfoundation.org>

From: Linus Torvalds <torvalds@linux-foundation.org>

commit b53119f13a04879c3bf502828d99d13726639ead upstream.

aio_poll() is not the only case that needs file pinned; worse, while
aio_read()/aio_write() can live without pinning iocb itself, the
proof is rather brittle and can easily break on later changes.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 fs/aio.c |   37 +++++++++++++++++++++----------------
 1 file changed, 21 insertions(+), 16 deletions(-)

--- a/fs/aio.c
+++ b/fs/aio.c
@@ -1022,6 +1022,9 @@ static bool get_reqs_available(struct ki
 /* aio_get_req
  *	Allocate a slot for an aio request.
  * Returns NULL if no requests are free.
+ *
+ * The refcount is initialized to 2 - one for the async op completion,
+ * one for the synchronous code that does this.
  */
 static inline struct aio_kiocb *aio_get_req(struct kioctx *ctx)
 {
@@ -1034,7 +1037,7 @@ static inline struct aio_kiocb *aio_get_
 	percpu_ref_get(&ctx->reqs);
 	req->ki_ctx = ctx;
 	INIT_LIST_HEAD(&req->ki_list);
-	refcount_set(&req->ki_refcnt, 0);
+	refcount_set(&req->ki_refcnt, 2);
 	req->ki_eventfd = NULL;
 	return req;
 }
@@ -1067,15 +1070,18 @@ out:
 	return ret;
 }
 
+static inline void iocb_destroy(struct aio_kiocb *iocb)
+{
+	if (iocb->ki_filp)
+		fput(iocb->ki_filp);
+	percpu_ref_put(&iocb->ki_ctx->reqs);
+	kmem_cache_free(kiocb_cachep, iocb);
+}
+
 static inline void iocb_put(struct aio_kiocb *iocb)
 {
-	if (refcount_read(&iocb->ki_refcnt) == 0 ||
-	    refcount_dec_and_test(&iocb->ki_refcnt)) {
-		if (iocb->ki_filp)
-			fput(iocb->ki_filp);
-		percpu_ref_put(&iocb->ki_ctx->reqs);
-		kmem_cache_free(kiocb_cachep, iocb);
-	}
+	if (refcount_dec_and_test(&iocb->ki_refcnt))
+		iocb_destroy(iocb);
 }
 
 static void aio_fill_event(struct io_event *ev, struct aio_kiocb *iocb,
@@ -1749,9 +1755,6 @@ static ssize_t aio_poll(struct aio_kiocb
 	INIT_LIST_HEAD(&req->wait.entry);
 	init_waitqueue_func_entry(&req->wait, aio_poll_wake);
 
-	/* one for removal from waitqueue, one for this function */
-	refcount_set(&aiocb->ki_refcnt, 2);
-
 	mask = vfs_poll(req->file, &apt.pt) & req->events;
 	if (unlikely(!req->head)) {
 		/* we did not manage to set up a waitqueue, done */
@@ -1782,7 +1785,6 @@ out:
 
 	if (mask)
 		aio_poll_complete(aiocb, mask);
-	iocb_put(aiocb);
 	return 0;
 }
 
@@ -1873,18 +1875,21 @@ static int __io_submit_one(struct kioctx
 		break;
 	}
 
+	/* Done with the synchronous reference */
+	iocb_put(req);
+
 	/*
 	 * If ret is 0, we'd either done aio_complete() ourselves or have
 	 * arranged for that to be done asynchronously.  Anything non-zero
 	 * means that we need to destroy req ourselves.
 	 */
-	if (ret)
-		goto out_put_req;
-	return 0;
+	if (!ret)
+		return 0;
+
 out_put_req:
 	if (req->ki_eventfd)
 		eventfd_ctx_put(req->ki_eventfd);
-	iocb_put(req);
+	iocb_destroy(req);
 out_put_reqs_available:
 	put_reqs_available(ctx, 1);
 	return ret;



  parent reply	other threads:[~2019-04-30 11:50 UTC|newest]

Thread overview: 106+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-30 11:37 [PATCH 5.0 00/89] 5.0.11-stable review Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 01/89] netfilter: nf_tables: bogus EBUSY when deleting set after flush Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 02/89] netfilter: nf_tables: bogus EBUSY in helper removal from transaction Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 03/89] intel_th: gth: Fix an off-by-one in output unassigning Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 04/89] powerpc/vdso32: fix CLOCK_MONOTONIC on PPC64 Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 05/89] ALSA: hda/realtek - Move to ACT_INIT state Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 06/89] fs/proc/proc_sysctl.c: Fix a NULL pointer dereference Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 07/89] block, bfq: fix use after free in bfq_bfqq_expire Greg Kroah-Hartman
2019-04-30 11:37 ` [PATCH 5.0 08/89] cifs: fix memory leak in SMB2_read Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 09/89] cifs: fix page reference leak with readv/writev Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 10/89] cifs: do not attempt cifs operation on smb2+ rename error Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 11/89] tracing: Fix a memory leak by early error exit in trace_pid_write() Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 12/89] tracing: Fix buffer_ref pipe ops Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 13/89] crypto: xts - Fix atomic sleep when walking skcipher Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 14/89] crypto: lrw " Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 15/89] gpio: eic: sprd: Fix incorrect irq type setting for the sync EIC Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 16/89] zram: pass down the bvec we need to read into in the work struct Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 17/89] lib/Kconfig.debug: fix build error without CONFIG_BLOCK Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 18/89] MIPS: scall64-o32: Fix indirect syscall number load Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 19/89] trace: Fix preempt_enable_no_resched() abuse Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 20/89] mm: do not boost watermarks to avoid fragmentation for the DISCONTIG memory model Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 21/89] arm64: mm: Ensure tail of unaligned initrd is reserved Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 22/89] IB/rdmavt: Fix frwr memory registration Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 23/89] RDMA/mlx5: Do not allow the user to write to the clock page Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 24/89] RDMA/mlx5: Use rdma_user_map_io for mapping BAR pages Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 25/89] RDMA/ucontext: Fix regression with disassociate Greg Kroah-Hartman
2019-05-03 11:47   ` Michal Kubecek
2019-05-03 11:48     ` Michal Kubecek
2019-04-30 11:38 ` [PATCH 5.0 26/89] sched/numa: Fix a possible divide-by-zero Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 27/89] ceph: only use d_name directly when parent is locked Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 28/89] ceph: ensure d_name stability in ceph_dentry_hash() Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 29/89] ceph: fix ci->i_head_snapc leak Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 30/89] nfsd: Dont release the callback slot unless it was actually held Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 31/89] nfsd: wake waiters blocked on file_lock before deleting it Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 32/89] nfsd: wake blocked file lock waiters before sending callback Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 33/89] sunrpc: dont mark uninitialised items as VALID Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 34/89] perf/x86/intel: Update KBL Package C-state events to also include PC8/PC9/PC10 counters Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 35/89] Input: synaptics-rmi4 - write config register values to the right offset Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 36/89] vfio/type1: Limit DMA mappings per container Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 37/89] dmaengine: sh: rcar-dmac: With cyclic DMA residue 0 is valid Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 38/89] dmaengine: sh: rcar-dmac: Fix glitch in dmaengine_tx_status Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 39/89] dmaengine: mediatek-cqdma: fix wrong register usage in mtk_cqdma_start Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 40/89] ARM: 8857/1: efi: enable CP15 DMB instructions before cleaning the cache Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 41/89] powerpc/mm/radix: Make Radix require HUGETLB_PAGE Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 42/89] drm/vc4: Fix memory leak during gpu reset Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 43/89] drm/ttm: fix re-init of global structures Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 44/89] Revert "drm/i915/fbdev: Actually configure untiled displays" Greg Kroah-Hartman
2019-05-01 13:02   ` Sasha Levin
2019-05-01 13:06     ` Sasha Levin
2019-05-01 13:08     ` Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 45/89] drm/vc4: Fix compilation error reported by kbuild test bot Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 46/89] USB: Add new USB LPM helpers Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 47/89] USB: Consolidate LPM checks to avoid enabling LPM twice Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 48/89] ext4: fix some error pointer dereferences Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 49/89] loop: do not print warn message if partition scan is successful Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 50/89] tipc: handle the err returned from cmd header function Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 51/89] slip: make slhc_free() silently accept an error pointer Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 52/89] workqueue: Try to catch flush_work() without INIT_WORK() Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 53/89] binder: fix handling of misaligned binder object Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 54/89] sched/deadline: Correctly handle active 0-lag timers Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 55/89] mac80211_hwsim: calculate if_combination.max_interfaces Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 56/89] NFS: Forbid setting AF_INET6 to "struct sockaddr_in"->sin_family Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 57/89] netfilter: ebtables: CONFIG_COMPAT: drop a bogus WARN_ON Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 58/89] fm10k: Fix a potential NULL pointer dereference Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 59/89] tipc: check bearer name with right length in tipc_nl_compat_bearer_enable Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 60/89] tipc: check link name with right length in tipc_nl_compat_link_set Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 61/89] net: netrom: Fix error cleanup path of nr_proto_init Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 62/89] net/rds: Check address length before reading address family Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 63/89] rxrpc: fix race condition in rxrpc_input_packet() Greg Kroah-Hartman
2019-04-30 11:38 ` Greg Kroah-Hartman [this message]
2019-04-30 11:38 ` [PATCH 5.0 65/89] aio: fold lookup_kiocb() into its sole caller Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 66/89] aio: keep io_event in aio_kiocb Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 67/89] aio: store event at final iocb_put() Greg Kroah-Hartman
2019-04-30 11:38 ` [PATCH 5.0 68/89] Fix aio_poll() races Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 69/89] x86, retpolines: Raise limit for generating indirect calls from switch-case Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 70/89] x86/retpolines: Disable switch jump tables when retpolines are enabled Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 71/89] rdma: fix build errors on s390 and MIPS due to bad ZERO_PAGE use Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 72/89] ipv4: add sanity checks in ipv4_link_failure() Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 73/89] ipv4: set the tcp_min_rtt_wlen range from 0 to one day Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 74/89] mlxsw: spectrum: Fix autoneg status in ethtool Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 75/89] net/mlx5e: ethtool, Remove unsupported SFP EEPROM high pages query Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 76/89] net: rds: exchange of 8K and 1M pool Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 77/89] net/rose: fix unbound loop in rose_loopback_timer() Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 78/89] net: stmmac: move stmmac_check_ether_addr() to driver probe Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 79/89] net/tls: fix refcount adjustment in fallback Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 80/89] stmmac: pci: Adjust IOT2000 matching Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 81/89] team: fix possible recursive locking when add slaves Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 82/89] net: socionext: replace napi_alloc_frag with the netdev variant on init Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 83/89] net/ncsi: handle overflow when incrementing mac address Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 84/89] mlxsw: pci: Reincrease PCI reset timeout Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 85/89] mlxsw: spectrum: Put MC TCs into DWRR mode Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 86/89] net/mlx5e: Fix the max MTU check in case of XDP Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 87/89] net/mlx5e: Fix use-after-free after xdp_return_frame Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 88/89] net/tls: avoid potential deadlock in tls_set_device_offload_rx() Greg Kroah-Hartman
2019-04-30 11:39 ` [PATCH 5.0 89/89] net/tls: dont leak IV and record seq when offload fails Greg Kroah-Hartman
2019-04-30 17:06 ` [PATCH 5.0 00/89] 5.0.11-stable review kernelci.org bot
2019-04-30 22:33 ` shuah
2019-05-01  7:55   ` Greg Kroah-Hartman
2019-05-01  6:21 ` Naresh Kamboju
2019-05-01  7:55   ` Greg Kroah-Hartman
2019-05-01  8:26 ` Jon Hunter
2019-05-01  8:43   ` Greg Kroah-Hartman
2019-05-01 16:44 ` Guenter Roeck
2019-05-01 17:14   ` Greg Kroah-Hartman
2019-05-02  5:30 ` Bharath Vedartham
2019-05-02  6:44   ` Greg Kroah-Hartman

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190430113612.718349457@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@roeck-us.net \
    --cc=stable@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=viro@zeniv.linux.org.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).