[5.17,101/114] writeback: Avoid skipping inode writeback

Message ID	20220516193628.376212768@linuxfoundation.org
State	Superseded
Headers	show Return-Path: <stable-owner@kernel.org> From: Greg Kroah-Hartman <gregkh@linuxfoundation.org> To: linux-kernel@vger.kernel.org Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>, stable@vger.kernel.org, Jan Kara <jack@suse.cz>, Christoph Hellwig <hch@lst.de>, Jing Xia <jing.xia@unisoc.com> Subject: [PATCH 5.17 101/114] writeback: Avoid skipping inode writeback Date: Mon, 16 May 2022 21:37:15 +0200 Message-Id: <20220516193628.376212768@linuxfoundation.org> In-Reply-To: <20220516193625.489108457@linuxfoundation.org> References: <20220516193625.489108457@linuxfoundation.org> User-Agent: quilt/0.66 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Precedence: bulk
Series	None \| expand [5.17,002/114] iwlwifi: iwl-dbg: Use del_timer_sync() before freeing [5.17,003/114] hwmon: (tmp401) Add OF device ID table [5.17,004/114] mac80211: Reset MBSSID parameters upon connection [5.17,005/114] net: rds: use maybe_get_net() when acquiring refcount on TCP sockets [5.17,006/114] net: Fix features skip in for_each_netdev_feature() [5.17,007/114] net: mscc: ocelot: fix last VCAP IS1/IS2 filter persisting in hardware when deleted [5.17,008/114] net: mscc: ocelot: fix VCAP IS2 filters matching on both lookups [5.17,009/114] net: mscc: ocelot: restrict tc-trap actions to VCAP IS2 lookup 0 [5.17,010/114] net: mscc: ocelot: avoid corrupting hardware counters when moving VCAP filters [5.17,011/114] fbdev: simplefb: Cleanup fb_info in .fb_destroy rather than .remove [5.17,012/114] fbdev: efifb: Cleanup fb_info in .fb_destroy rather than .remove [5.17,013/114] fbdev: vesafb: Cleanup fb_info in .fb_destroy rather than .remove [5.17,014/114] platform/surface: aggregator: Fix initialization order when compiling as builtin m... [5.17,015/114] ice: Fix race during aux device (un)plugging [5.17,016/114] ice: clear stale Tx queue settings before configuring [5.17,017/114] ice: fix PTP stale Tx timestamps cleanup [5.17,018/114] ipv4: drop dst in multicast routing path [5.17,019/114] drm/nouveau: Fix a potential theorical leak in nouveau_get_backlight_name() [5.17,020/114] netlink: do not reset transport header in netlink_recvmsg() [5.17,021/114] net: chelsio: cxgb4: Avoid potential negative array offset [5.17,022/114] fbdev: efifb: Fix a use-after-free due early fb_info cleanup [5.17,023/114] net: sfc: fix memory leak due to ptp channel [5.17,024/114] fanotify: do not allow setting dirent events in mask of non-dir [5.17,025/114] mac80211_hwsim: call ieee80211_tx_prepare_skb under RCU protection [5.17,026/114] nfs: fix broken handling of the softreval mount option [5.17,027/114] ionic: fix missing pci_release_regions() on error in ionic_probe() [5.17,028/114] dim: initialize all struct fields [5.17,029/114] hwmon: (ltq-cputemp) restrict it to SOC_XWAY [5.17,030/114] procfs: prevent unprivileged processes accessing fdinfo dir [5.17,031/114] selftests: vm: Makefile: rename TARGETS to VMTARGETS [5.17,032/114] net: dsa: flush switchdev workqueue on bridge join error path [5.17,033/114] arm64: vdso: fix makefile dependency on vdso.so [5.17,034/114] virtio: fix virtio transitional ids [5.17,035/114] s390/ctcm: fix variable dereferenced before check [5.17,036/114] s390/ctcm: fix potential memory leak [5.17,037/114] s390/lcs: fix variable dereferenced before check [5.17,038/114] net/sched: act_pedit: really ensure the skb is writable [5.17,039/114] net: ethernet: mediatek: ppe: fix wrong size passed to memset() [5.17,040/114] net: bcmgenet: Check for Wake-on-LAN interrupt probe deferral [5.17,041/114] drm/vc4: hdmi: Fix build error for implicit function declaration [5.17,042/114] mlxsw: Avoid warning during ip6gre device removal [5.17,043/114] net: dsa: bcm_sf2: Fix Wake-on-LAN with mac_link_down() [5.17,044/114] net/smc: non blocking recvmsg() return -EAGAIN when no data and signal_pending [5.17,045/114] net: sfc: ef10: fix memory leak in efx_ef10_mtd_probe() [5.17,046/114] tls: Fix context leak on tls_device_down [5.17,047/114] drm/vmwgfx: Fix fencing on SVGAv3 [5.17,048/114] gfs2: Fix filesystem block deallocation for short writes [5.17,049/114] hwmon: (asus_wmi_sensors) Fix CROSSHAIR VI HERO name [5.17,050/114] hwmon: (f71882fg) Fix negative temperature [5.17,051/114] RDMA/irdma: Fix deadlock in irdma_cleanup_cm_core() [5.17,052/114] iommu: arm-smmu: disable large page mappings for Nvidia arm-smmu [5.17,053/114] ASoC: max98090: Reject invalid values in custom control put() [5.17,054/114] ASoC: max98090: Generate notifications on changes for custom control [5.17,055/114] ASoC: ops: Validate input values in snd_soc_put_volsw_range() [5.17,056/114] s390: disable -Warray-bounds [5.17,057/114] ASoC: SOF: Fix NULL pointer exception in sof_pci_probe callback [5.17,058/114] io_uring: assign non-fixed early for async work [5.17,059/114] net: emaclite: Dont advertise 1000BASE-T and do auto negotiation [5.17,060/114] net: sfp: Add tx-fault workaround for Huawei MA5671A SFP ONT [5.17,061/114] secure_seq: use the 64 bits of the siphash for port offset calculation [5.17,062/114] tcp: use different parts of the port_offset for index and offset [5.17,063/114] tcp: resalt the secret every 10 seconds [5.17,064/114] tcp: add small random increments to the source port [5.17,065/114] tcp: dynamically allocate the perturb table used by source ports [5.17,066/114] tcp: increase source port perturb table to 2^16 [5.17,067/114] tcp: drop the hash_32() part from the index calculation [5.17,068/114] block: Do not call folio_next() on an unreferenced folio [5.17,069/114] interconnect: Restore sync state by ignoring ipa-virt in provider count [5.17,070/114] perf tests: Fix coresight `perf test` failure. [5.17,071/114] firmware_loader: use kernel credentials when reading firmware [5.17,072/114] KVM: PPC: Book3S PR: Enable MSR_DR for switch_mmu_context() [5.17,073/114] usb: xhci-mtk: fix fs isocs transfer error [5.17,074/114] x86/mm: Fix marking of unused sub-pmd ranges [5.17,075/114] tty/serial: digicolor: fix possible null-ptr-deref in digicolor_uart_probe() [5.17,076/114] tty: n_gsm: fix buffer over-read in gsm_dlci_data() [5.17,077/114] tty: n_gsm: fix mux activation issues in gsm_config() [5.17,078/114] tty: n_gsm: fix invalid gsmtty_write_room() result [5.17,079/114] usb: gadget: uvc: allow for application to cleanly shutdown [5.17,080/114] usb: cdc-wdm: fix reading stuck on device close [5.17,081/114] usb: typec: tcpci: Dont skip cleanup in .remove() on error [5.17,082/114] usb: typec: tcpci_mt6360: Update for BMC PHY setting [5.17,083/114] USB: serial: pl2303: add device id for HP LM930 Display [5.17,084/114] USB: serial: qcserial: add support for Sierra Wireless EM7590 [5.17,085/114] USB: serial: option: add Fibocom L610 modem [5.17,086/114] USB: serial: option: add Fibocom MA510 modem [5.17,087/114] slimbus: qcom: Fix IRQ check in qcom_slim_probe [5.17,088/114] fsl_lpuart: Dont enable interrupts too early [5.17,089/114] genirq: Remove WARN_ON_ONCE() in generic_handle_domain_irq() [5.17,090/114] serial: 8250_mtk: Fix UART_EFR register address [5.17,091/114] serial: 8250_mtk: Fix register address for XON/XOFF character [5.17,092/114] ceph: fix setting of xattrs on async created inodes [5.17,093/114] Revert "mm/memory-failure.c: skip huge_zero_page in memory_failure()" [5.17,094/114] mm/huge_memory: do not overkill when splitting huge_zero_page [5.17,095/114] mm: mremap: fix sign for EFAULT error return value [5.17,096/114] drm/vmwgfx: Disable command buffers on svga3 without gbobjects [5.17,097/114] drm/nouveau/tegra: Stop using iommu_present() [5.17,098/114] i40e: i40e_main: fix a missing check on list iterator [5.17,099/114] net: atlantic: always deep reset on pm op, fixing up my null deref regression [5.17,100/114] net: phy: Fix race condition on link status change [5.17,101/114] writeback: Avoid skipping inode writeback [5.17,102/114] cgroup/cpuset: Remove cpus_allowed/mems_allowed setup in cpuset_init_smp() [5.17,103/114] ping: fix address binding wrt vrf [5.17,104/114] ath11k: reduce the wait time of 11d scan and hw scan while add interface [5.17,105/114] arm[64]/memremap: dont abuse pfn_valid() to ensure presence of linear map [5.17,106/114] net: phy: micrel: Do not use kszphy_suspend/resume for KSZ8061 [5.17,107/114] net: phy: micrel: Pass .probe for KS8737 [5.17,108/114] SUNRPC: Ensure that the gssproxy client can start in a connected state [5.17,109/114] drm/vmwgfx: Initialize drm_mode_fb_cmd2 [5.17,110/114] Revert "drm/amd/pm: keep the BACO feature enabled for suspend" [5.17,111/114] dma-buf: call dma_buf_stats_setup after dmabuf is in valid list [5.17,112/114] mm/hwpoison: use pr_err() instead of dump_page() in get_any_page() [5.17,113/114] net: phy: micrel: Fix incorrect variable type in micrel [5.17,114/114] mm/kfence: reset PG_slab and memcg_data before freeing __kfence_pool

Message ID

20220516193628.376212768@linuxfoundation.org

State

Superseded

Headers

From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        stable@vger.kernel.org, Jan Kara <jack@suse.cz>,
        Christoph Hellwig <hch@lst.de>, Jing Xia <jing.xia@unisoc.com>
Subject: [PATCH 5.17 101/114] writeback: Avoid skipping inode writeback
Date: Mon, 16 May 2022 21:37:15 +0200
Message-Id: <20220516193628.376212768@linuxfoundation.org>
In-Reply-To: <20220516193625.489108457@linuxfoundation.org>
References: <20220516193625.489108457@linuxfoundation.org>
User-Agent: quilt/0.66
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

Series

None | expand

Commit Message

Greg Kroah-Hartman May 16, 2022, 7:37 p.m. UTC

From: Jing Xia <jing.xia@unisoc.com>

commit 846a3351ddfe4a86eede4bb26a205c3f38ef84d3 upstream.

We have run into an issue that a task gets stuck in
balance_dirty_pages_ratelimited() when perform I/O stress testing.
The reason we observed is that an I_DIRTY_PAGES inode with lots
of dirty pages is in b_dirty_time list and standard background
writeback cannot writeback the inode.
After studing the relevant code, the following scenario may lead
to the issue:

task1                                   task2
-----                                   -----
fuse_flush
 write_inode_now //in b_dirty_time
  writeback_single_inode
   __writeback_single_inode
                                 fuse_write_end
                                  filemap_dirty_folio
                                   __xa_set_mark:PAGECACHE_TAG_DIRTY
    lock inode->i_lock
    if mapping tagged PAGECACHE_TAG_DIRTY
    inode->i_state |= I_DIRTY_PAGES
    unlock inode->i_lock
                                   __mark_inode_dirty:I_DIRTY_PAGES
                                      lock inode->i_lock
                                      -was dirty,inode stays in
                                      -b_dirty_time
                                      unlock inode->i_lock

   if(!(inode->i_state & I_DIRTY_All))
      -not true,so nothing done

This patch moves the dirty inode to b_dirty list when the inode
currently is not queued in b_io or b_more_io list at the end of
writeback_single_inode.

Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
CC: stable@vger.kernel.org
Fixes: 0ae45f63d4ef ("vfs: add support for a lazytime mount option")
Signed-off-by: Jing Xia <jing.xia@unisoc.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220510023514.27399-1-jing.xia@unisoc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 fs/fs-writeback.c |    4 ++++
 1 file changed, 4 insertions(+)

--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -1749,6 +1749,10 @@  static int writeback_single_inode(struct
 	 */
 	if (!(inode->i_state & I_DIRTY_ALL))
 		inode_cgwb_move_to_attached(inode, wb);
+	else if (!(inode->i_state & I_SYNC_QUEUED) &&
+		 (inode->i_state & I_DIRTY))
+		redirty_tail_locked(inode, wb);
+
 	spin_unlock(&wb->list_lock);
 	inode_sync_complete(inode);
 out:

[5.17,101/114] writeback: Avoid skipping inode writeback

Commit Message

Patch