btrfs: add preferred_metadata mode (with last btrfs_find_device argument, without mount option)
When this mode is enabled, the allocation policy of the chunk
is so modified:
- allocation of metadata chunk: priority is given to preferred_metadata
disks.
- allocation of data chunk: priority is given to a non preferred_metadata
disk.
When a striped profile is involved (like RAID0,5,6), the logic
is a bit more complex. If there are enough disks, the data profiles
are stored on the non preferred_metadata disks; instead the metadata
profiles are stored on the preferred_metadata disk.
If the disks are not enough, then the profile is allocated on all
the disks.
Example: assuming that sda, sdb, sdc are ssd disks, and sde, sdf are
non preferred_metadata ones.
A data profile raid6, will be stored on sda, sdb, sdc, sde, sdf (sde
and sdf are not enough to host a raid5 profile).
A metadata profile raid6, will be stored on sda, sdb, sdc (these
are enough to host a raid6 profile).
To enable this mode pass -o dedicated_metadata at mount time.
Driver ->power_on and ->power_off callbacks leaks internal SMCC firmware
return codes to phy caller. This patch converts SMCC error codes to
standard linux errno codes. Include file linux/arm-smccc.h already provides
defines for SMCC error codes, so use them instead of custom driver defines.
Note that return value is signed 32bit, but stored in unsigned long type
with zero padding.
Tested-by: Tomasz Maciej Nowak <tmn505@gmail.com> Link: https://lore.kernel.org/r/20200902144344.16684-2-pali@kernel.org Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reviewed-by: Rob Herring <robh@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A build failure was raised by kbuild with the following error.
drivers/android/binder.c: Assembler messages:
drivers/android/binder.c:3861: Error: unrecognized keyword/register name `l.lwz ?ap,4(r24)'
drivers/android/binder.c:3866: Error: unrecognized keyword/register name `l.addi ?ap,r0,0'
The issue is with 64-bit get_user() calls on openrisc. I traced this to
a problem where in the internally in the get_user macros there is a cast
to long __gu_val this causes GCC to think the get_user call is 32-bit.
This binder code is really long and GCC allocates register r30, which
triggers the issue. The 64-bit get_user asm tries to get the 64-bit pair
register, which for r30 overflows the general register names and returns
the dummy register ?ap.
The fix here is to move the temporary variables into the asm macros. We
use a 32-bit __gu_tmp for 32-bit and smaller macro and a 64-bit tmp in
the 64-bit macro. The cast in the 64-bit macro has a trick of casting
through __typeof__((x)-(x)) which avoids the below warning. This was
barrowed from riscv.
arch/openrisc/include/asm/uaccess.h:240:8: warning: cast to pointer from integer of different size
I tested this in a small unit test to check reading between 64-bit and
32-bit pointers to 64-bit and 32-bit values in all combinations. Also I
ran make C=1 to confirm no new sparse warnings came up. It all looks
clean to me.
There seems to be a bug in the original code when gntdev_get_page()
is called with writeable=true then the page needs to be marked dirty
before being put.
To address this, a bool writeable is added in gnt_dev_copy_batch, set
it in gntdev_grant_copy_seg() (and drop `writeable` argument to
gntdev_get_page()) and then, based on batch->writeable, use
set_page_dirty_lock().
Fixes: a4cdb556cae0 (xen/gntdev: add ioctl for grant copy) Suggested-by: Boris Ostrovsky <boris.ostrovsky@oracle.com> Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com> Cc: Juergen Gross <jgross@suse.com> Cc: David Vrabel <david.vrabel@citrix.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/1599375114-32360-1-git-send-email-jrdr.linux@gmail.com Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Before commit 9495b7e92f716ab2 ("driver core: platform: Initialize
dma_parms for platform devices"), the R-Car SATA device didn't have DMA
parameters. Hence the DMA boundary mask supplied by its driver was
silently ignored, as __scsi_init_queue() doesn't check the return value
of dma_set_seg_boundary(), and the default value of 0xffffffff was used.
Now the device has gained DMA parameters, the driver-supplied value is
used, and the following warning is printed on Salvator-XS:
DMA-API: sata_rcar ee300000.sata: mapping sg segment across boundary [start=0x00000000ffffe000] [end=0x00000000ffffefff] [boundary=0x000000001ffffffe]
WARNING: CPU: 5 PID: 38 at kernel/dma/debug.c:1233 debug_dma_map_sg+0x298/0x300
(the range of start/end values depend on whether IOMMU support is
enabled or not)
The issue here is that SATA_RCAR_DMA_BOUNDARY doesn't have bit 0 set, so
any typical end value, which is odd, will trigger the check.
Fix this by increasing the DMA boundary value by 1.
This also fixes the following WRITE DMA EXT timeout issue:
Commit 8234f6734c5d ("PM-runtime: Switch autosuspend over to using
hrtimers") switched PM runtime autosuspend to use hrtimers and all
related time accounting in ns, but missed to update the timer_expires
data type in struct dev_pm_info to u64.
This causes the timer_expires value to be truncated on 32-bit
architectures when assignment is done from u64 values:
The issue being that CPU0 takes 'port_lock' on the irq path in pl011_int()
before taking 'console_owner' on the printk() path, whereas CPU1 takes
the two locks in the opposite order on the printk() path due to setting
the "console_owner" prior to calling into into the actual console driver.
Fix this in the same way as the msm-serial driver by dropping 'port_lock'
before handling the sysrq.
For QUP IP versions 2.5 and above the oversampling rate is
halved from 32 to 16.
Commit ce734600545f ("tty: serial: qcom_geni_serial: Update
the oversampling rate") is pushed to handle this scenario.
But the existing logic is failing to classify QUP Version 3.0
into the correct group ( 2.5 and above).
As result Serial Engine clocks are not configured properly for
baud rate and garbage data is sampled to FIFOs from the line.
So, fix the logic to detect QUP with versions 2.5 and above.
Since the debugfs may peek into the GEM contexts as the corresponding
client/fd is being closed, we may try and follow a dangling pointer.
However, the context closure itself is serialised with the ctx->mutex,
so if we hold that mutex as we inspect the state coupled in the context,
we know the pointers within the context are stable and will remain valid
as we inspect their tables.
process_one_req() // again
req->callback() // BOOM
cancel_delayed_work_sync()
The solution is to always cancel the work once it is completed so any
in between set_timeout() does not result in it running again.
Cc: stable@vger.kernel.org Fixes: 44e75052bc2a ("RDMA/rdma_cm: Make rdma_addr_cancel into a fence") Link: https://lore.kernel.org/r/20200930072007.1009692-1-leon@kernel.org Reported-by: Dan Aloni <dan@kernelim.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In p54p_tx(), skb->data is mapped to streaming DMA on line 337:
mapping = pci_map_single(..., skb->data, ...);
Then skb->data is accessed on line 349:
desc->device_addr = ((struct p54_hdr *)skb->data)->req_id;
This access may cause data inconsistency between CPU cache and hardware.
To fix this problem, ((struct p54_hdr *)skb->data)->req_id is stored in
a local variable before DMA mapping, and then the driver accesses this
local variable instead of skb->data.
Cc: <stable@vger.kernel.org> Signed-off-by: Jia-Ju Bai <baijiaju@tsinghua.edu.cn> Acked-by: Christian Lamparter <chunkeey@gmail.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20200802132949.26788-1-baijiaju@tsinghua.edu.cn Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch checks the size for the EVM_IMA_XATTR_DIGSIG and
EVM_XATTR_PORTABLE_DIGSIG types to ensure that the algorithm is read from
the buffer returned by vfs_getxattr_alloc().
Older ATF does not provide SMC call for SATA phy power on functionality and
therefore initialization of ahci_mvebu is failing when older version of ATF
is using. In this case phy_power_on() function returns -EOPNOTSUPP.
This patch adds a new hflag AHCI_HFLAG_IGN_NOTSUPP_POWER_ON which cause
that ahci_platform_enable_phys() would ignore -EOPNOTSUPP errors from
phy_power_on() call.
It fixes initialization of ahci_mvebu on Espressobin boards where is older
Marvell's Arm Trusted Firmware without SMC call for SATA phy power.
This is regression introduced in commit 8e18c8e58da64 ("arm64: dts: marvell:
armada-3720-espressobin: declare SATA PHY property") where SATA phy was
defined and therefore ahci_platform_enable_phys() on Espressobin started
failing.
Stephane Eranian found a bug in that IBS' current Fetch counter was not
being reset when the driver would write the new value to clear it along
with the enable bit set, and found that adding an MSR write that would
first disable IBS Fetch would make IBS Fetch reset its current count.
Indeed, the PPR for AMD Family 17h Model 31h B0 55803 Rev 0.54 - Sep 12,
2019 states "The periodic fetch counter is set to IbsFetchCnt [...] when
IbsFetchEn is changed from 0 to 1."
Explicitly set IbsFetchEn to 0 and then to 1 when re-enabling IBS Fetch,
so the driver properly resets the internal counter to 0 and IBS
Fetch starts counting again.
A family 15h machine tested does not have this problem, and the extra
wrmsr is also not needed on Family 19h, so only do the extra wrmsr on
families 16h through 18h.
Don't recheck it since xattr_permission() already
checks CAP_SYS_ADMIN capability.
Just follow 5d3ce4f70172 ("f2fs: avoid duplicated permission check for "trusted." xattrs")
Reported-by: Hongyu Jin <hongyu.jin@unisoc.com>
[ Gao Xiang: since it could cause some complex Android overlay
permission issue as well on android-5.4+, it'd be better to
backport to 5.4+ rather than pure cleanup on mainline. ] Cc: <stable@vger.kernel.org> # 5.4+ Link: https://lore.kernel.org/r/20200811070020.6339-1-hsiangkao@redhat.com Reviewed-by: Chao Yu <yuchao0@huawei.com> Signed-off-by: Gao Xiang <hsiangkao@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The tcf_block_unbind() expects that the caller will take block->cb_lock
before calling it, however the code took RTNL lock and dropped cb_lock
instead. This causes to the following kernel panic.
Commit ed42989eab57 ("tipc: fix the skb_unshare() in tipc_buf_append()")
replaced skb_unshare() with skb_copy() to not reduce the data reference
counter of the original skb intentionally. This is not the correct
way to handle the cloned skb because it causes memory leak in 2
following cases:
1/ Sending multicast messages via broadcast link
The original skb list is cloned to the local skb list for local
destination. After that, the data reference counter of each skb
in the original list has the value of 2. This causes each skb not
to be freed after receiving ACK:
tipc_link_advance_transmq()
{
...
/* release skb */
__skb_unlink(skb, &l->transmq);
kfree_skb(skb); <-- memory exists after being freed
}
2/ Sending multicast messages via replicast link
Similar to the above case, each skb cannot be freed after purging
the skb list:
tipc_mcast_xmit()
{
...
__skb_queue_purge(pkts); <-- memory exists after being freed
}
This commit fixes this issue by using skb_unshare() instead. Besides,
to avoid use-after-free error reported by KASAN, the pointer to the
fragment is set to NULL before calling skb_unshare() to make sure that
the original skb is not freed after freeing the fragment 2 times in
case skb_unshare() returns NULL.
Fixes: ed42989eab57 ("tipc: fix the skb_unshare() in tipc_buf_append()") Acked-by: Jon Maloy <jmaloy@redhat.com> Reported-by: Thang Hoang Ngo <thang.h.ngo@dektech.com.au> Signed-off-by: Tung Nguyen <tung.q.nguyen@dektech.com.au> Reviewed-by: Xin Long <lucien.xin@gmail.com> Acked-by: Cong Wang <xiyou.wangcong@gmail.com> Link: https://lore.kernel.org/r/20201027032403.1823-1-tung.q.nguyen@dektech.com.au Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With SO_RCVLOWAT, under memory pressure,
it is possible to enter a state where:
1. We have not received enough bytes to satisfy SO_RCVLOWAT.
2. We have not entered buffer pressure (see tcp_rmem_pressure()).
3. But, we do not have enough buffer space to accept more packets.
In this case, we advertise 0 rwnd (due to #3) but the application does
not drain the receive queue (no wakeup because of #1 and #2) so the
flow stalls.
Modify the heuristic for SO_RCVLOWAT so that, if we are advertising
rwnd<=rcv_mss, force a wakeup to prevent a stall.
Without this patch, setting tcp_rmem to 6143 and disabling TCP
autotune causes a stalled flow. With this patch, no stall occurs. This
is with RPC-style traffic with large messages.
Fixes: 03f45c883c6f ("tcp: avoid extra wakeups for SO_RCVLOWAT users") Signed-off-by: Arjun Roy <arjunroy@google.com> Acked-by: Soheil Hassas Yeganeh <soheil@google.com> Acked-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Link: https://lore.kernel.org/r/20201023184709.217614-1-arjunroy.kdev@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the function ravb_hwtstamp_get() in ravb_main.c with the existing
values for RAVB_RXTSTAMP_TYPE_V2_L2_EVENT (0x2) and RAVB_RXTSTAMP_TYPE_ALL
(0x6)
if (priv->tstamp_rx_ctrl & RAVB_RXTSTAMP_TYPE_V2_L2_EVENT)
config.rx_filter = HWTSTAMP_FILTER_PTP_V2_L2_EVENT;
else if (priv->tstamp_rx_ctrl & RAVB_RXTSTAMP_TYPE_ALL)
config.rx_filter = HWTSTAMP_FILTER_ALL;
if the test on RAVB_RXTSTAMP_TYPE_ALL should be true,
it will never be reached.
This issue can be verified with 'hwtstamp_config' testing program
(tools/testing/selftests/net/hwtstamp_config.c). Setting filter type
to ALL and subsequent retrieving it gives incorrect value:
$ hwtstamp_config eth0 OFF ALL
flags = 0
tx_type = OFF
rx_filter = ALL
$ hwtstamp_config eth0
flags = 0
tx_type = OFF
rx_filter = PTP_V2_L2_EVENT
Correct this by converting if-else's to switch.
Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper") Reported-by: Julia Lawall <julia.lawall@inria.fr> Signed-off-by: Andrew Gabbasov <andrew_gabbasov@mentor.com> Reviewed-by: Sergei Shtylyov <sergei.shtylyov@gmail.com> Link: https://lore.kernel.org/r/20201026102130.29368-1-andrew_gabbasov@mentor.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As reported by Serge flag IRQF_NO_THREAD causes an error if the
interrupt is actually shared and the other driver(s) don't have this
flag set. This situation can occur if a PCI(e) legacy interrupt is
used in combination with forced threading.
There's no good way to deal with this properly, therefore we have to
remove flag IRQF_NO_THREAD. For fixing the original forced threading
issue switch to napi_schedule().
IPA transactions describe actions to be performed by the IPA
hardware. Three cases use IPA transactions: transmitting a socket
buffer; providing a page to receive packet data; and issuing an IPA
immediate command. An IPA transaction contains a scatter/gather
list (SGL) to hold the set of actions to be performed.
We map buffers in the SGL for DMA at the time they are added to the
transaction. For skb TX transactions, we fill the SGL with a call
to skb_to_sgvec(). Page RX transactions involve a single page
pointer, and that is recorded in the SGL with sg_set_page(). In
both of these cases we then map the SGL for DMA with a call to
dma_map_sg().
Immediate commands are different. The payload for an immediate
command comes from a region of coherent DMA memory, which must
*not* be mapped for DMA. For that reason, gsi_trans_cmd_add()
sort of hand-crafts each SGL entry added to a command transaction.
This patch fixes a problem with the code that crafts the SGL entry
for an immediate command. Previously a portion of the SGL entry was
updated using sg_set_buf(). However this is not valid because it
includes a call to virt_to_page() on the buffer, but the command
buffer pointer is not a linear address.
Since we never actually map the SGL for command transactions, there
are very few fields in the SGL we need to fill. Specifically, we
only need to record the DMA address and the length, so they can be
used by __gsi_trans_commit() to fill a TRE. We additionally need to
preserve the SGL flags so for_each_sg() still works. For that we
can simply assign a null page pointer for command SGL entries.
Fixes: 9dd441e4ed575 ("soc: qcom: ipa: GSI transactions") Reported-by: Stephen Boyd <swboyd@chromium.org> Tested-by: Stephen Boyd <swboyd@chromium.org> Signed-off-by: Alex Elder <elder@linaro.org> Link: https://lore.kernel.org/r/20201022010029.11877-1-elder@linaro.org Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It looks like the BAR memory region had already been unmapped before we
start clearing CMDQ registers in it, which is pretty bad and the kernel
happily kills itself because of a Current EL Data Abort (on arm64).
Moving the CMDQ uninitialization a bit early fixes the issue for me.
Fixes: 862d969a3a4d ("net: hns3: do VF's pci re-initialization while PF doing FLR") Signed-off-by: Zenghui Yu <yuzenghui@huawei.com> Link: https://lore.kernel.org/r/20201023051550.793-1-yuzenghui@huawei.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently it is possible to craft a special netlink RTM_NEWQDISC
command that can result in jitter being equal to 0x80000000. It is
enough to set the 32 bit jitter to 0x02000000 (it will later be
multiplied by 2^6) or just set the 64 bit jitter via
TCA_NETEM_JITTER64. This causes an overflow during the generation of
uniformly distributed numbers in tabledist(), which in turn leads to
division by zero (sigma != 0, but sigma * 2 is 0).
The related fragment of code needs 32-bit division - see commit 9b0ed89 ("netem: remove unnecessary 64 bit modulus"), so switching to
64 bit is not an option.
Fix the issue by keeping the value of jitter within the range that can
be adequately handled by tabledist() - [0;INT_MAX]. As negative std
deviation makes no sense, take the absolute value of the passed value
and cap it at INT_MAX. Inside tabledist(), switch to unsigned 32 bit
arithmetic in order to prevent overflows.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Aleksandr Nogikh <nogikh@google.com> Reported-by: syzbot+ec762a6342ad0d3c0d8f@syzkaller.appspotmail.com Acked-by: Stephen Hemminger <stephen@networkplumber.org> Link: https://lore.kernel.org/r/20201028170731.1383332-1-aleksandrnogikh@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jakub Kicinski brought up a concern in ibmvnic_set_mac().
ibmvnic_set_mac() does this:
ether_addr_copy(adapter->mac_addr, addr->sa_data);
if (adapter->state != VNIC_PROBED)
rc = __ibmvnic_set_mac(netdev, addr->sa_data);
So if state == VNIC_PROBED, the user can assign an invalid address to
adapter->mac_addr, and ibmvnic_set_mac() will still return 0.
The fix is to validate ethernet address at the beginning of
ibmvnic_set_mac(), and move the ether_addr_copy to
the case of "adapter->state != VNIC_PROBED".
Fixes: c26eba03e407 ("ibmvnic: Update reset infrastructure to support tunable parameters") Signed-off-by: Lijun Pan <ljp@linux.ibm.com> Link: https://lore.kernel.org/r/20201027220456.71450-1-ljp@linux.ibm.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The check for src mac address in ibmveth_is_packet_unsupported is wrong.
Commit 6f2275433a2f wanted to shut down messages for loopback packets,
but now suppresses bridged frames, which are accepted by the hypervisor
otherwise bridging won't work at all.
Fixes: 6f2275433a2f ("ibmveth: Detect unsupported packets before sending to the hypervisor") Signed-off-by: Michal Suchanek <msuchanek@suse.de> Signed-off-by: Thomas Bogendoerfer <tbogendoerfer@suse.de> Link: https://lore.kernel.org/r/20201026104221.26570-1-msuchanek@suse.de Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current code sets up the filter action field before
rewrites are set up. When the action 'switch' is used
with rewrites, this may result in initial few packets
that get switched out don't have rewrites applied
on them.
So, make sure filter action is set up along with rewrites
or only after everything else is set up for rewrites.
Fixes: 12b276fbf6e0 ("cxgb4: add support to create hash filters") Signed-off-by: Raju Rangoju <rajur@chelsio.com> Link: https://lore.kernel.org/r/20201023115852.18262-1-rajur@chelsio.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
chtls_pt_recvmsg() receives a skb with tls header and subsequent
skb with data, need to finalize the data copy whenever next skb
with tls header is available. but here current tls header is
overwritten by next available tls header, ends up corrupting
user buffer data. fixing it by finalizing current record whenever
next skb contains tls header.
v1->v2:
- Improved commit message.
Fixes: 17a7d24aa89d ("crypto: chtls - generic handling of data and hdr") Signed-off-by: Vinay Kumar Yadav <vinay.yadav@chelsio.com> Link: https://lore.kernel.org/r/20201022190556.21308-1-vinay.yadav@chelsio.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
CPL handler functions chtls_pass_open_rpl() and
chtls_close_listsrv_rpl() should return CPL_RET_BUF_DONE
so that caller function will do skb free to avoid leak.
In chtls_pass_establish() we hold child socket lock using bh_lock_sock
and we are again trying bh_lock_sock in add_to_reap_list, causing deadlock.
Remove bh_lock_sock in add_to_reap_list() as lock is already held.
In the AER or firmware reset flow, if we are in fatal error state or
if pci_channel_offline() is true, we don't send any commands to the
firmware because the commands will likely not reach the firmware and
most commands don't matter much because the firmware is likely to be
reset imminently.
However, the HWRM_FUNC_RESET command is different and we should always
attempt to send it. In the AER flow for example, the .slot_reset()
call will trigger this fw command and we need to try to send it to
effect the proper reset.
Fixes: b340dc680ed4 ("bnxt_en: Avoid sending firmware messages when AER error is detected.") Reviewed-by: Edwin Peer <edwin.peer@broadcom.com> Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When a PCIe fatal error occurs, the internal latched BAR addresses
in the chip get reset even though the BAR register values in config
space are retained.
pci_restore_state() will not rewrite the BAR addresses if the
BAR address values are valid, causing the chip's internal BAR addresses
to stay invalid. So we need to zero the BAR registers during PCIe fatal
error to force pci_restore_state() to restore the BAR addresses. These
write cycles to the BAR registers will cause the proper BAR addresses to
latch internally.
As part of the commit b148bb238c02
("bnxt_en: Fix possible crash in bnxt_fw_reset_task()."),
cancel_delayed_work_sync() is called only for VFs to fix a possible
crash by cancelling any pending delayed work items. It was assumed
by mistake that the flush_workqueue() call on the PF would flush
delayed work items as well.
As flush_workqueue() does not cancel the delayed workqueue, extend
the fix for PFs. This fix will avoid the system crash, if there are
any pending delayed work items in fw_reset_task() during driver's
.remove() call.
Unify the workqueue cleanup logic for both PF and VF by calling
cancel_work_sync() and cancel_delayed_work_sync() directly in
bnxt_remove_one().
Fixes: b148bb238c02 ("bnxt_en: Fix possible crash in bnxt_fw_reset_task().") Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Reviewed-by: Andy Gospodarek <gospo@broadcom.com> Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A recent patch has moved the workqueue cleanup logic before
calling unregister_netdev() in bnxt_remove_one(). This caused a
regression because the workqueue can be restarted if the device is
still open. Workqueue cleanup must be done after unregister_netdev().
The workqueue will not restart itself after the device is closed.
Call bnxt_cancel_sp_work() after unregister_netdev() and
call bnxt_dl_fw_reporters_destroy() after that. This fixes the
regession and the original NULL ptr dereference issue.
Fixes: b16939b59cc0 ("bnxt_en: Fix NULL ptr dereference crash in bnxt_fw_reset_task()") Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
bnxt_open_nic() is called during configuration changes that require
the NIC to be closed and then opened. This call is protected by
rtnl_lock. Firmware reset can be happening at the same time. Only
critical portions of the entire firmware reset sequence are protected
by the rtnl_lock. It is possible that bnxt_open_nic() can be called
when the firmware reset sequence is aborting. In that case,
bnxt_open_nic() needs to check if the ABORT_ERR flag is set and
abort if it is. The configuration change that resulted in the
bnxt_open_nic() call will fail but the NIC will be brought to a
consistent IF_DOWN state.
Without this patch, if bnxt_open_nic() were to continue in this error
state, it may crash like this:
Without this patch efivarfs_alloc_dentry creates dentries with slashes in
their name if the respective EFI variable has slashes in its name. This in
turn causes EIO on getdents64, which prevents a complete directory listing
of /sys/firmware/efi/efivars/.
This patch replaces the invalid shlashes with exclamation marks like
kobject_set_name_vargs does for /sys/firmware/efi/vars/ to have consistently
named dentries under /sys/firmware/efi/vars/ and /sys/firmware/efi/efivars/.
Signed-off-by: Michael Schaller <misch@google.com> Link: https://lore.kernel.org/r/20200925074502.150448-1-misch@google.com Signed-off-by: Ard Biesheuvel <ardb@kernel.org> Signed-off-by: dann frazier <dann.frazier@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The motivations to go rework memcpy_mcsafe() are that the benefit of
doing slow and careful copies is obviated on newer CPUs, and that the
current opt-in list of CPUs to instrument recovery is broken relative to
those CPUs. There is no need to keep an opt-in list up to date on an
ongoing basis if pmem/dax operations are instrumented for recovery by
default. With recovery enabled by default the old "mcsafe_key" opt-in to
careful copying can be made a "fragile" opt-out. Where the "fragile"
list takes steps to not consume poison across cachelines.
The discussion with Linus made clear that the current "_mcsafe" suffix
was imprecise to a fault. The operations that are needed by pmem/dax are
to copy from a source address that might throw #MC to a destination that
may write-fault, if it is a user page.
So copy_to_user_mcsafe() becomes copy_mc_to_user() to indicate
the separate precautions taken on source and destination.
copy_mc_to_kernel() is introduced as a non-SMAP version that does not
expect write-faults on the destination, but is still prepared to abort
with an error code upon taking #MC.
The original copy_mc_fragile() implementation had negative performance
implications since it did not use the fast-string instruction sequence
to perform copies. For this reason copy_mc_to_kernel() fell back to
plain memcpy() to preserve performance on platforms that did not indicate
the capability to recover from machine check exceptions. However, that
capability detection was not architectural and now that some platforms
can recover from fast-string consumption of memory errors the memcpy()
fallback now causes these more capable platforms to fail.
Introduce copy_mc_enhanced_fast_string() as the fast default
implementation of copy_mc_to_kernel() and finalize the transition of
copy_mc_fragile() to be a platform quirk to indicate 'copy-carefully'.
With this in place, copy_mc_to_kernel() is fast and recovery-ready by
default regardless of hardware capability.
Thanks to Vivek for identifying that copy_user_generic() is not suitable
as the copy_mc_to_user() backend since the #MC handler explicitly checks
ex_has_fault_handler(). Thanks to the 0day robot for catching a
performance bug in the x86/copy_mc_to_user implementation.
[ bp: Add the "why" for this change from the 0/2th message, massage. ]
In reaction to a proposal to introduce a memcpy_mcsafe_fast()
implementation Linus points out that memcpy_mcsafe() is poorly named
relative to communicating the scope of the interface. Specifically what
addresses are valid to pass as source, destination, and what faults /
exceptions are handled.
Of particular concern is that even though x86 might be able to handle
the semantics of copy_mc_to_user() with its common copy_user_generic()
implementation other archs likely need / want an explicit path for this
case:
On Fri, May 1, 2020 at 11:28 AM Linus Torvalds <torvalds@linux-foundation.org> wrote:
>
> On Thu, Apr 30, 2020 at 6:21 PM Dan Williams <dan.j.williams@intel.com> wrote:
> >
> > However now I see that copy_user_generic() works for the wrong reason.
> > It works because the exception on the source address due to poison
> > looks no different than a write fault on the user address to the
> > caller, it's still just a short copy. So it makes copy_to_user() work
> > for the wrong reason relative to the name.
>
> Right.
>
> And it won't work that way on other architectures. On x86, we have a
> generic function that can take faults on either side, and we use it
> for both cases (and for the "in_user" case too), but that's an
> artifact of the architecture oddity.
>
> In fact, it's probably wrong even on x86 - because it can hide bugs -
> but writing those things is painful enough that everybody prefers
> having just one function.
Replace a single top-level memcpy_mcsafe() with either
copy_mc_to_user(), or copy_mc_to_kernel().
Introduce an x86 copy_mc_fragile() name as the rename for the
low-level x86 implementation formerly named memcpy_mcsafe(). It is used
as the slow / careful backend that is supplanted by a fast
copy_mc_generic() in a follow-on patch.
One side-effect of this reorganization is that separating copy_mc_64.S
to its own file means that perf no longer needs to track dependencies
for its memcpy_64.S benchmarks.
Fix build error when CONFIG_ACPI is not set/enabled by adding the header
file <asm/acpi.h> which contains a stub for the function in the build
error.
../arch/x86/pci/intel_mid_pci.c: In function ‘intel_mid_pci_init’:
../arch/x86/pci/intel_mid_pci.c:303:2: error: implicit declaration of function ‘acpi_noirq_set’; did you mean ‘acpi_irq_get’? [-Werror=implicit-function-declaration]
acpi_noirq_set();
Fixes: a912a7584ec3 ("x86/platform/intel-mid: Move PCI initialization to arch_init()") Link: https://lore.kernel.org/r/ea903917-e51b-4cc9-2680-bc1e36efa026@infradead.org Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> Reviewed-by: Jesse Barnes <jsbarnes@google.com> Acked-by: Thomas Gleixner <tglx@linutronix.de> Cc: stable@vger.kernel.org # v4.16+ Cc: Jacob Pan <jacob.jun.pan@linux.intel.com> Cc: Len Brown <lenb@kernel.org> Cc: Jesse Barnes <jsbarnes@google.com> Cc: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With CONFIG_EXPERT=y, CONFIG_KASAN=y, CONFIG_RANDOMIZE_BASE=n,
CONFIG_RELOCATABLE=n, we observe the following failure when trying to
link the kernel image with LD=ld.lld:
error: section: .exit.data is not contiguous with other relro sections
ld.lld defaults to -z relro while ld.bfd defaults to -z norelro. This
was previously fixed, but only for CONFIG_RELOCATABLE=y.
Fixes: 3bbd3db86470 ("arm64: relocatable: fix inconsistencies in linker script and options") Signed-off-by: Nick Desaulniers <ndesaulniers@google.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20201016175339.2429280-1-ndesaulniers@google.com Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 606f8e7b27bf ("arm64: capabilities: Use linear array for
detection and verification") changed the way we deal with per-CPU errata
by only calling the .matches() callback until one CPU is found to be
affected. At this point, .matches() stop being called, and .cpu_enable()
will be called on all CPUs.
This breaks the ARCH_WORKAROUND_2 handling, as only a single CPU will be
mitigated.
In order to address this, forcefully call the .matches() callback from a
.cpu_enable() callback, which brings us back to the original behaviour.
Fixes: 606f8e7b27bf ("arm64: capabilities: Use linear array for detection and verification") Cc: <stable@vger.kernel.org> Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Marc Zyngier <maz@kernel.org> Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 73f381660959 ("arm64: Advertise mitigation of Spectre-v2, or lack
thereof") changed the way we deal with ARCH_WORKAROUND_1, by moving most
of the enabling code to the .matches() callback.
This has the unfortunate effect that the workaround gets only enabled on
the first affected CPU, and no other.
In order to address this, forcefully call the .matches() callback from a
.cpu_enable() callback, which brings us back to the original behaviour.
Fixes: 73f381660959 ("arm64: Advertise mitigation of Spectre-v2, or lack thereof") Cc: <stable@vger.kernel.org> Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Marc Zyngier <maz@kernel.org> Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The "FIRMWARE_EFI_EMBEDDED" enum is a "where", not a "what". It
should not be distinguished separately from just "FIRMWARE", as this
confuses the LSMs about what is being loaded. Additionally, there was
no actual validation of the firmware contents happening.
Fixes: e4c2c0ff00ec ("firmware: Add new platform fallback mechanism and firmware_request_platform()") Signed-off-by: Kees Cook <keescook@chromium.org> Reviewed-by: Luis Chamberlain <mcgrof@kernel.org> Acked-by: Scott Branden <scott.branden@broadcom.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20201002173828.2099543-3-keescook@chromium.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently, on arm64, we abort on any failure from efi_get_random_bytes()
other than EFI_NOT_FOUND when it comes to setting the physical seed for
KASLR, but ignore such failures when obtaining the seed for virtual
KASLR or for early seeding of the kernel's entropy pool via the config
table. This is inconsistent, and may lead to unexpected boot failures.
So let's permit any failure for the physical seed, and simply report
the error code if it does not equal EFI_NOT_FOUND.
When building for an embedded target using Yocto, we're sometimes
observing that the version string that gets built into vmlinux (and
thus what uname -a reports) differs from the path under /lib/modules/
where modules get installed in the rootfs, but only in the length of
the -gabc123def suffix. Hence modprobe always fails.
The problem is that Yocto has the concept of "sstate" (shared state),
which allows different developers/buildbots/etc. to share build
artifacts, based on a hash of all the metadata that went into building
that artifact - and that metadata includes all dependencies (e.g. the
compiler used etc.). That normally works quite well; usually a clean
build (without using any sstate cache) done by one developer ends up
being binary identical to a build done on another host. However, one
thing that can cause two developers to end up with different builds
[and thus make one's vmlinux package incompatible with the other's
kernel-dev package], which is not captured by the metadata hashing, is
this `git describe`: The output of that can be affected by
(1) git version: before 2.11 git defaulted to a minimum of 7, since
2.11 (git.git commit e6c587) the default is dynamic based on the
number of objects in the repo
(2) hence even if both run the same git version, the output can differ
based on how many remotes are being tracked (or just lots of local
development branches or plain old garbage)
(3) and of course somebody could have a core.abbrev config setting in
~/.gitconfig
So in order to avoid `uname -a` output relying on such random details
of the build environment which are rather hard to ensure are
consistent between developers and buildbots, make sure the abbreviated
sha1 always consists of exactly 12 hex characters. That is consistent
with the current rule for -stable patches, and is almost always enough
to identify the head commit unambigously - in the few cases where it
does not, the v5.4.3-00021- prefix would certainly nail it down.
There are no bugs here that I've spotted, it's just easier to use the
normal API and there are no performance advantages to using the more
verbose advanced API.
The xas_store() wasn't paired with an xas_nomem() loop, so if it couldn't
allocate memory using GFP_NOWAIT, it would leak the reference to the file
descriptor. Also the node pointed to by the xas could be freed between
the call to xas_load() under the rcu_read_lock() and the acquisition of
the xa_lock.
It's easier to just use the normal xa_load/xa_store interface here.
which is complaining about xa_destroy() grabbing the xa lock in an
IRQ disabling fashion, whereas the io_uring uses cases aren't interrupt
safe. This is really an xarray issue, since it should not assume the
lock type. But for our use case, since we know the xarray is empty at
this point, there's no need to actually call xa_destroy(). So just get
rid of it.
/* all workers gone, wq exit can proceed */
if (!nr_workers && refcount_dec_and_test(&wqe->wq->refs))
complete(&wqe->wq->done);
because there might be multiple cases of wqe in a wq and we would wait
for every worker in every wqe to go home before releasing wq's resources
on destroying.
To that end, rework wq's refcount by making it independent of the tracking
of workers because after all they are two different things, and keeping
it balanced when workers come and go. Note the manager kthread, like
other workers, now holds a grab to wq during its lifetime.
Finally to help destroy wq, check IO_WQ_BIT_EXIT upon creating worker
and do nothing for exiting wq.
During a context switch the scheduler invokes wq_worker_sleeping() with
disabled preemption. Disabling preemption is needed because it protects
access to `worker->sleeping'. As an optimisation it avoids invoking
schedule() within the schedule path as part of possible wake up (thus
preempt_enable_no_resched() afterwards).
The io-wq has been added to the mix in the same section with disabled
preemption. This breaks on PREEMPT_RT because io_wq_worker_sleeping()
acquires a spinlock_t. Also within the schedule() the spinlock_t must be
acquired after tsk_is_pi_blocked() otherwise it will block on the
sleeping lock again while scheduling out.
While playing with `io_uring-bench' I didn't notice a significant
latency spike after converting io_wqe::lock to a raw_spinlock_t. The
latency was more or less the same.
In order to keep the spinlock_t it would have to be moved after the
tsk_is_pi_blocked() check which would introduce a branch instruction
into the hot path.
The lock is used to maintain the `work_list' and wakes one task up at
most.
Should io_wqe_cancel_pending_work() cause latency spikes, while
searching for a specific item, then it would need to drop the lock
during iterations.
revert_creds() is also invoked under the lock. According to debug
cred::non_rcu is 0. Otherwise it should be moved outside of the locked
section because put_cred_rcu()->free_uid() acquires a sleeping lock.
Convert io_wqe::lock to a raw_spinlock_t.c
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we don't get and assign the namespace for the async work, then certain
paths just don't work properly (like /dev/stdin, /proc/mounts, etc).
Anything that references the current namespace of the given task should
be assigned for async work on behalf of that task.
Grab actual references to the files_struct. To avoid circular references
issues due to this, we add a per-task note that keeps track of what
io_uring contexts a task has used. When the tasks execs or exits its
assigned files, we cancel requests based on this tracking.
With that, we can grab proper references to the files table, and no
longer need to rely on stashing away ring_fd and ring_file to check
if the ring_fd may have been closed.
Sometimes we assign a weak reference to it, sometimes we grab a
reference to it. Clean this up and make it unconditional, and drop the
flag related to tracking this state.
We can grab a reference to the task instead of stashing away the task
files_struct. This is doable without creating a circular reference
between the ring fd and the task itself.
We currently cancel these when the ring exits, and we cancel all of
them. This is in preparation for killing only the ones associated
with a given task.
This isn't safe, and isn't needed either. We are guaranteed that any
work we queue is on a live task (and will be run), or it goes to
our backup io-wq threads if the task is exiting.
BUG: KASAN: slab-out-of-bounds in nft_flow_rule_create+0x622/0x6a2
net/netfilter/nf_tables_offload.c:40
Read of size 8 at addr ffff888103910b58 by task syz-executor227/16244
The error happens when expr->ops is accessed early on before performing the boundary check and after nft_expr_next() moves the expr to go out-of-bounds.
This patch checks the boundary condition before expr->ops that fixes the slab-out-of-bounds Read issue.
Add nft_expr_more() and use it to fix this problem.
Currently, enabling f_ncm at SuperSpeed Plus speeds results in an
oops in config_ep_by_speed because ncm_set_alt passes in NULL
ssp_descriptors. Fix this by re-using the SuperSpeed descriptors.
This is safe because usb_assign_descriptors calls
usb_copy_descriptors.
Tested: enabled f_ncm on a dwc3 gadget and 10Gbps link, ran iperf Reviewed-by: Maciej Żenczykowski <maze@google.com> Signed-off-by: Lorenzo Colitti <lorenzo@google.com> Signed-off-by: Felipe Balbi <balbi@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / linux / log