| Age | Commit message (Collapse) | Author |
|---|
|
Commit fb586f25300f ("sctp: delay calls to sk_data_ready() as much as
possible") minimized the number of wake ups that are triggered in case
the association receives a packet with multiple data chunks on it and/or
when io_events are enabled and then commit 0970f5b36659 ("sctp: signal
sk_data_ready earlier on data chunks reception") moved the wake up to as
soon as possible. It thus relies on the state machine running later to
clean the flag that the event was already generated.
The issue is that there are 2 call paths that calls
sctp_ulpq_tail_event() outside of the state machine, causing the flag to
linger and possibly omitting a needed wake up in the sequence.
One of the call paths is when enabling SCTP_SENDER_DRY_EVENTS via
setsockopt(SCTP_EVENTS), as noticed by Harald Welte. The other is when
partial reliability triggers removal of chunks from the send queue when
the application calls sendmsg().
This commit fixes it by not setting the flag in case the socket is not
owned by the user, as it won't be cleaned later. This works for
user-initiated calls and also for rx path processing.
Fixes: fb586f25300f ("sctp: delay calls to sk_data_ready() as much as possible")
Reported-by: Harald Welte <laforge@gnumonks.org>
Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
The netlink message sent with type == 0, which doesn't have any client
behind it, caused to the overflow in max_num_ops array.
Fix it by declaring zero number of ops for the first client.
Fixes: c9901724a2f1 ("RDMA/netlink: Remove netlink clients infrastructure")
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
64bit division causes build/link errors on 32bit architectures. It
prints out error messages like:
ERROR: "__aeabi_uldivmod" [net/netfilter/xt_hashlimit.ko] undefined!
The value of avg passed through by userspace in BYTE mode cannot exceed
U32_MAX. Which means 64bit division in user2rate_bytes is unnecessary.
To fix this I have changed the type of param 'user' to u32.
Since anything greater than U32_MAX is an invalid input we error out in
hashlimit_mt_check_common() when this is the case.
Changes in v2:
Making return type as u32 would cause an overflow for small
values of 'user' (for example 2, 3 etc). To avoid this I bumped up
'r' to u64 again as well as the return type. This is OK since the
variable that stores the result is u64. We still avoid 64bit
division here since 'user' is u32.
Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
Signed-off-by: Vishwanath Pai <vpai@akamai.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
struct xt_byteslimit_htable used hlist_head, but memory allocation is
done through sizeof(struct list_head).
Signed-off-by: Zhizhou Tian <zhizhou.tian@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
kernel test robot reported:
WARNING: CPU: 0 PID: 1244 at net/netfilter/core.c:218 __nf_hook_entries_try_shrink+0x49/0xcd
[..]
After allowing batching in nf_unregister_net_hooks its possible that an earlier
call to __nf_hook_entries_try_shrink already compacted the list.
If this happens we don't need to do anything.
Fixes: d3ad2c17b4047 ("netfilter: core: batch nf_unregister_net_hooks synchronize_net calls")
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Aaron Conole <aconole@bytheb.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
no need to serialize on a single lock, we can partition the table and
add/delete in parallel to different slots.
This restores one of the advantages that got lost with the rhlist
revert.
Cc: Ivan Babrou <ibobrik@gmail.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
This reverts commit 870190a9ec9075205c0fa795a09fa931694a3ff1.
It was not a good idea. The custom hash table was a much better
fit for this purpose.
A fast lookup is not essential, in fact for most cases there is no lookup
at all because original tuple is not taken and can be used as-is.
What needs to be fast is insertion and deletion.
rhlist removal however requires a rhlist walk.
We can have thousands of entries in such a list if source port/addresses
are reused for multiple flows, if this happens removal requests are so
expensive that deletions of a few thousand flows can take several
seconds(!).
The advantages that we got from rhashtable are:
1) table auto-sizing
2) multiple locks
1) would be nice to have, but it is not essential as we have at
most one lookup per new flow, so even a million flows in the bysource
table are not a problem compared to current deletion cost.
2) is easy to add to custom hash table.
I tried to add hlist_node to rhlist to speed up rhltable_remove but this
isn't doable without changing semantics. rhltable_remove_fast will
check that the to-be-deleted object is part of the table and that
requires a list walk that we want to avoid.
Furthermore, using hlist_node increases size of struct rhlist_head, which
in turn increases nf_conn size.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=196821
Reported-by: Ivan Babrou <ibobrik@gmail.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
There are reports about spurious softlockups during iptables-restore, a
backtrace i saw points at get_counters -- it uses a sequence lock and also
has unbounded restart loop.
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
It seems preferrable to limp along if we have a conflicting mapping,
its certainly better than a BUG().
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
fib6_net_exit only frees the main and local tables. If another table was
created with fib6_alloc_table, we leak it when the netns is destroyed.
Fix this in the same way ip_fib_net_exit cleans up tables, by walking
through the whole hashtable of fib6_table's. We can get rid of the
special cases for local and main, since they're also part of the
hashtable.
Reproducer:
ip netns add x
ip -net x -6 rule add from 6003:1::/64 table 100
ip netns del x
Reported-by: Jianlin Shi <jishi@redhat.com>
Fixes: 58f09b78b730 ("[NETNS][IPV6] ip6_fib - make it per network namespace")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
Now that we have the DRM_DEV_* variants, we should use them.
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170908143218.9701-1-seanpaul@chromium.org
|
|
This commit applies memory-barriers.txt part of upstream change, commit
706eeb3e9c6f ("Documentation/locking/atomic: Add documents for new
atomic_t APIs") to Korean translation.
Signed-off-by: SeongJae Park <sj38.park@gmail.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
|
|
This commit applies upstream change, commit 66ce3a4dcb9f ("doc: Update
memory-barriers.txt for read-to-write dependencies") to Korean
translation.
Signed-off-by: SeongJae Park <sj38.park@gmail.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
|
|
Only the media PDF book was requiring adjustbox, in order to
scale big tables. That worked pretty good with Sphinx versions
1.4 and 1.5, but Spinx 1.6 changed the way tables are produced,
by introducing some weird macros before tabulary.
That causes adjustbox to fail. So, it can't be used anymore,
and its usage was removed from the media book.
So, let's remove it from conf.py and sphinx-pre-install.
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
|
|
Sphinx 1.5 added a new way to change backward colors for note
boxes, but kept backward compatibility with 1.4. On Sphinx 1.6,
the old way stopped working, in favor of a new less hackish
way.
Unfortunately, this is currently too buggy to be used, and
the old way doesn't work anymore. So, we have no option but
to stick with boring notice boxes.
One example of such bug is the notice that it is inside
struct v4l2_plane, at the "bytesused" field.
At least, add a notice about how to use, as maybe some day
the bug will vanish.
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
|
|
Only the media book used this extension in the past, but
it is not required anymore.
Cleanup patch only.
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
|
|
KASAN reported the following:
[ 19.338655] ==================================================================
[ 19.345946] BUG: KASAN: use-after-free in of_platform_device_destroy+0x88/0x100
[ 19.345966] Read of size 8 at addr fffffe01aa6f1468 by task systemd-udevd/264
[ 19.345983] CPU: 1 PID: 264 Comm: systemd-udevd Not tainted 4.13.0-jang+ #737
[ 19.345989] Hardware name: Cavium ThunderX CN81XX board (DT)
[ 19.345995] Call trace:
[ 19.346013] [<fffffc800808b1b0>] dump_backtrace+0x0/0x368
[ 19.346026] [<fffffc800808b6bc>] show_stack+0x24/0x30
[ 19.346040] [<fffffc8008cbb944>] dump_stack+0xa4/0xc8
[ 19.346057] [<fffffc80082c2870>] print_address_description+0x68/0x258
[ 19.346070] [<fffffc80082c2d70>] kasan_report+0x238/0x2f8
[ 19.346082] [<fffffc80082c14a8>] __asan_load8+0x88/0xb8
[ 19.346098] [<fffffc8008aacee0>] of_platform_device_destroy+0x88/0x100
[ 19.346131] [<fffffc8000e02fa4>] thunder_mmc_probe+0x314/0x550 [thunderx_mmc]
[ 19.346147] [<fffffc800879d560>] pci_device_probe+0x158/0x1f8
[ 19.346162] [<fffffc800886e53c>] driver_probe_device+0x394/0x5f8
[ 19.346174] [<fffffc800886e8f4>] __driver_attach+0x154/0x158
[ 19.346185] [<fffffc800886b12c>] bus_for_each_dev+0xdc/0x140
[ 19.346196] [<fffffc800886d9f8>] driver_attach+0x38/0x48
[ 19.346207] [<fffffc800886d148>] bus_add_driver+0x290/0x3c8
[ 19.346219] [<fffffc800886fc5c>] driver_register+0xbc/0x1a0
[ 19.346232] [<fffffc800879b78c>] __pci_register_driver+0xc4/0xd8
[ 19.346260] [<fffffc8000e80024>] thunder_mmc_driver_init+0x24/0x10000 [thunderx_mmc]
[ 19.346273] [<fffffc8008083a80>] do_one_initcall+0x98/0x1c0
[ 19.346289] [<fffffc8008177b54>] do_init_module+0xe0/0x2cc
[ 19.346303] [<fffffc8008175cf0>] load_module+0x3238/0x35c0
[ 19.346318] [<fffffc8008176438>] SyS_finit_module+0x190/0x1a0
[ 19.346329] [<fffffc80080834a0>] __sys_trace_return+0x0/0x4
This is caused by:
platform_device_register()
-> platform_device_unregister(to_platform_device(dev))
freeing struct device
-> of_node_clear_flag(dev->of_node, ...)
writing to the freed device
The issue is solved by increasing the reference count before calling
of_platform_device_destroy() so freeing the device is postponed after
the call.
Fixes: 8fb83b142823 ("mmc: cavium: Fix probing race with regulator")
Signed-off-by: Jan Glauber <jglauber@cavium.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
|
|
MMC_DEBUG was moved and one letter got strangely capitalized.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
|
|
mmc_init_request() depends on card->bouncesz so it must be calculated
before blk_init_allocated_queue() starts allocating requests.
Reported-by: Seraphime Kirkovski <kirkseraph@gmail.com>
Fixes: 304419d8a7e9 ("mmc: core: Allocate per-request data using the..")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Tested-by: Seraphime Kirkovski <kirkseraph@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Tested-by: Pavel Machek <pavel@ucw.cz>
|
|
Use new blk-mq interfaces. Use multiple queues and also use the block
layer complete helper that finish the IO on the CPU that initiated it.
Reviewed-by: Jan Hoeppner <hoeppner@linux.vnet.ibm.com>
Signed-off-by: Stefan Haberland <sth@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
This fix was intended for 4.13, but didn't get in because both
maintainers were on vacation.
Paul Mackerras:
"It adds mutual exclusion between list_add_rcu and list_del_rcu calls
on the kvm->arch.spapr_tce_tables list. Without this, userspace could
potentially trigger corruption of the list and cause a host crash or
worse."
|
|
There's no reason for ipvs to create a conn for an ABORT packet
even if sysctl_sloppy_sctp is set.
This patch is to accept it without creating a conn, just as ipvs
does for tcp's RST packet.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
Commit 5e26b1b3abce ("ipvs: support scheduling inverse and icmp SCTP
packets") changed to check packet type early. It introduced a side
effect: if it's not a INIT packet, ports will be set as NULL, and
the packet will be dropped later.
It caused that sctp couldn't create connection when ipvs module is
loaded and any scheduler is registered on server.
Li Shuang reproduced it by running the cmds on sctp server:
# ipvsadm -A -t 1.1.1.1:80 -s rr
# ipvsadm -D -t 1.1.1.1:80
then the server could't work any more.
This patch is to return 1 when it's not an INIT packet. It means ipvs
will accept it without creating a conn for it, just like what it does
for tcp.
Fixes: 5e26b1b3abce ("ipvs: support scheduling inverse and icmp SCTP packets")
Reported-by: Li Shuang <shuali@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
|
|
The firmware feature check introduced for multi-scheduled scan is also
failing for bcm4345 devices resulting in a firmware crash.
The reason for this crash has not yet been root cause so this patch avoids
the feature check for those device as a short-term fix.
Fixes: 9fe929aaace6 ("brcmfmac: add firmware feature detection for gscan feature")
Cc: <stable@vger.kernel.org> # v4.13
Signed-off-by: Ian W MORRISON <ianwmorrison@gmail.com>
Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
|
|
Now that the last users have been converted, we can finally get rid of
for_each_obj_in_state, we have better macros to replace them with.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: David Airlie <airlied@linux.ie>
Link: https://patchwork.freedesktop.org/patch/msgid/20170719143920.25685-8-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
|
|
By always keeping track of the last commit in plane_state, we know
whether there is an active update on the plane or not. With that
information we can reject the fast update, and force the slowpath
to be used as was originally intended.
We cannot use plane_state->crtc->state here, because this only mentions
the most recent commit for the crtc, but not the planes that were part
of it. We specifically care about what the last commit involving this
plane is, which can only be tracked with a pointer in the plane state.
Changes since v1:
- Clean up the whole function here, instead of partially earlier.
- Add mention in the commit message why we need commit in plane_state.
- Swap plane->state in intel_legacy_cursor_update, instead of
reassigning all variables. With this commit We know that the cursor
is not part of any active commits so this hack can be removed.
Cc: Gustavo Padovan <gustavo.padovan@collabora.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> #v1
Link: https://patchwork.freedesktop.org/patch/msgid/20170904104838.23822-7-maarten.lankhorst@linux.intel.com
[mlankhorst: Amend commit for merge conflicts with drm-intel]
|
|
Currently we neatly track the crtc state, but forget to look at
plane/connector state.
When doing a nonblocking modeset, immediately followed by a setprop
before the modeset completes, the setprop will see the modesets new
state as the old state and free it.
This has to be solved by waiting for hw_done on the connector, even
if it's not assigned to a crtc. When a connector is unbound we take
the last crtc commit, and when it stays unbound we create a new
fake crtc commit for that gets signaled on hw_done for all the
planes/connectors.
We wait for it the same way as we do for crtc's, which will make
sure we never run into a use-after-free situation.
Changes since v1:
- Only create a single disable commit. (danvet)
- Fix leak in intel_legacy_cursor_update.
Changes since v2:
- Make reference counting in drm_atomic_helper_setup_commit
more obvious. (pinchartl)
- Call cleanup_done for fake commit. (danvet)
- Add comments to drm_atomic_helper_setup_commit. (danvet, pinchartl)
- Add comment to drm_atomic_helper_swap_state. (pinchartl)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Testcase: kms_atomic_transition.plane-use-after-nonblocking-unbind*
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170904104838.23822-6-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
|
|
When commit synchronization through drm_crtc_commit was first
introduced, we tried to solve the problem of the flip_done
needing a reference count by blocking in cleanup_done.
This has been changed by commit 24835e442f28 ("drm: reference count
event->completion") which made the waits here no longer needed.
However, even after this commit we still needed the wait because
otherwise we cannot wait for the flip_done because this item might
have been removed from the list.
Changed since v1:
- Make mention of cleanup_done completing before flip_done.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170904104838.23822-4-maarten.lankhorst@linux.intel.com
|
|
Most code only cares about the current commit or previous commit.
Fortuantely we already have a place to track those. Move it to
drm_crtc_state where it belongs. :)
The per-crtc commit_list is kept for places where we have to look
deeper than the current or previous commit for checking whether to stall
on unpin. This is used in drm_atomic_helper_setup_commit and
intel_has_pending_fb_unpin.
Changes since v1:
- Update kerneldoc for drm_crtc.commit_list. (danvet)
Changes since v2:
- Remove drm_atomic_helper_async_check hunk. (pinchartl)
Changes since v3:
- Fix use-after-free in drm_atomic_helper_commit_cleanup_done().
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170904150456.31049-1-maarten.lankhorst@linux.intel.com
[mlankhorst: preceeding -> preceding (checkpatch)]
|
|
This will allow code to do x->commit = drm_crtc_commit_get(commit),
making it clearer where references are used.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170904104838.23822-5-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
|
|
The next commit removes the wait for flip_done in in
drm_atomic_helper_commit_cleanup_done, but we need it for the tests
to pass. Instead of using complicated vblank tracking which ends
up being ignored anyway, call the correct atomic helper. :)
Changes since v1:
- Always call drm_atomic_helper_wait_for_flip_done,
even for legacy cursor updates. (danvet)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170904104838.23822-2-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
|
|
gcc-7 complains about multiplying within a condition being
suspicious:
drivers/gpu/drm/stm/dw_mipi_dsi-stm.c: In function 'dsi_pll_get_clkout_khz':
drivers/gpu/drm/stm/dw_mipi_dsi-stm.c:117:10: error: '*' in boolean context, suggest '&&' instead [-Werror=int-in-bool-context]
The code here is correct, but can be easily rephrased to make
that more obvious. I also swap out the error handling and the normal
code path for clarity.
Fixes: b0f09a3c69d9 ("drm/stm: Add STM32 DSI controller driver")
Acked-by: Philippe Cornu <philippe.cornu@st.com>
Tested-by: Philippe Cornu <philippe.cornu@st.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170906131331.2691300-1-arnd@arndb.de
|
|
drm-intel-next-queued
gvt-next-2017-09-08
- PCI config sanitize series (Changbin)
- Workload submission error handling series (Fred)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170908063155.l54lvpivxntjm7hq@zhen-hp.sh.intel.com
|
|
This adds support for Rockchip soc lvds found on rk3288
Based on the patches from Mark yao and Heiko Stuebner.
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sandy Huang <hjc@rock-chips.com>
Reviewed-by: Mark Yao <mark.yao@rock-chips.com>
Tested-by: Wadim Egorov <w.egorov@phytec.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1504351737-136042-1-git-send-email-hjc@rock-chips.com
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>
|
|
This patch add Document for Rockchip Soc RK3288 LVDS,
This based on the patches from Mark yao and Heiko Stuebner.
Signed-off-by: Mark yao <mark.yao@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sandy Huang <hjc@rock-chips.com>
Reviewed-by: Mark Yao <mark.yao@rock-chips.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/1504351729-135932-1-git-send-email-hjc@rock-chips.com
|
|
IGD is PCIe device and has extended configuration space. Checking
the binary dump, we can see we have Caps located out of PCI compatible
Configuration Space range.
0x000: 86 80 12 19 17 04 10 00 06 00 00 03 00 00 00 00
0x010: 04 00 00 10 08 00 00 00 0c 00 00 00 08 00 00 00
0x020: 00 00 00 00 00 00 00 00 00 00 00 00 28 10 b9 06
0x030: 00 f8 ff ff 40 00 00 00 00 00 00 00 0b 01 00 00
0x040: 09 70 0c 01 71 26 01 62 c8 00 04 84 00 00 00 00
0x050: c1 00 00 00 39 00 00 00 00 00 00 00 01 00 00 a2
0x060: 00 00 01 00 00 00 00 00 00 00 00 00 00 00 00 00
0x070: 10 ac 92 00 00 80 00 10 00 00 00 00 00 00 00 00
0x080: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x090: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0a0: 00 00 00 00 00 00 00 00 00 00 00 00 05 d0 01 00
0x0b0: 18 00 e0 fe 00 00 00 00 00 00 00 00 00 00 00 00
0x0c0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0d0: 01 00 22 00 00 80 00 00 00 00 00 00 00 00 00 00
0x0e0: 00 00 00 00 00 00 00 00 00 80 00 00 00 00 00 00
0x0f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x100: 1b 00 01 20 02 14 00 00 00 00 00 00 00 00 00 00
...
Currently, we only emulate the PCI compatible Configuration Space.
This is okay if we attach vGPU to PCI bus. But when we attach to
a PCI Express bus (when Qemu emulates a Intel Q35 chipset which has
PCIe slot), it will not work. Extended Configuration Space is required
for a PCIe device.
This patch extended the virtual configuration space from 256 bytes
to 4KB bytes. So we are to be a *real* PCIe device. And for the
Extended CapList we keep same to physical GPU.
Cc: Laszlo Ersek <lersek@redhat.com>
Tested-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Changbin Du <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
Looking at our virtual PCI device, we can see surprising Region 4 and Region 5.
00:10.0 VGA compatible controller: Intel Corporation Sky Lake Integrated Graphics (rev 06) (prog-if 00 [VGA controller])
....
Region 0: Memory at 140000000 (64-bit, non-prefetchable) [size=16M]
Region 2: Memory at 180000000 (64-bit, prefetchable) [size=1G]
Region 4: Memory at <ignored> (32-bit, non-prefetchable)
Region 5: Memory at <ignored> (32-bit, non-prefetchable)
Expansion ROM at febd6000 [disabled] [size=2K]
The fact is that we only implemented BAR0 and BAR2. Surprising Region 4 and
Region 5 are shown because we report their size as 0xffffffff. They should
report size 0 instead.
BTW, the physical GPU has a PIO BAR. GVTg hasn't implemented PIO access, so
we ignored this BAR for vGPU device.
v2: fix BAR size value calculation.
Link: https://bugzilla.redhat.com/show_bug.cgi?id=1458032
Signed-off-by: Changbin Du <changbin.du@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
fix the wrong return type and return error once the unknown
command is scanned.
v2:
- separate this error handle from healthy rating code. (Zhenyu)
Signed-off-by: fred gao <fred.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
When an error occurs in dispatch_workload, this patch is to do the
proper cleanup and rollback to the original states before the workload
is abandoned.
v2:
- split the mixed several error paths for better review. (Zhenyu)
v3:
- original PTR_ERR(cs) is good and code cleanup. (Zhenyu)
v4:
- reuse the existing i915_add_request for error handling. (Zhenyu)
v5:
- remove the duplicate error handling release_shadow_wa_ctx and
move the engine->context_unpin upper. (Zhenyu)
v6:
- keep the old label "out". (Zhenyu)
Signed-off-by: fred gao <fred.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
When it is failed in shadow_mm, the pin_count should rollback
to the original states before return.
v2:
- split the mixed several error paths for better review. (Zhenyu)
v3:
increase the pincount after shadow success. (Zhenyu)
Signed-off-by: fred gao <fred.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
refine the error handling for prepare_execlist_workload to restore to the
original states once error occurs.
only release the shadowed batch buffer and wa ctx when the workload is
completed successfully.
v2:
- split the mixed several error paths for better review. (Zhenyu)
v3:
- handle prepare batch buffer/wa ctx pin errors and
- emulate_schedule_in null issue. (Zhenyu)
v4:
- no need to handle emulate_schedule_in null issue. (Zhenyu)
v5:
- release the shadowed batch buffer and wa ctx only for the
successful workload. (Zhenyu)
v6:
- polish the return style. (Zhenyu)
Signed-off-by: fred gao <fred.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
When an error occurs after shadow_indirect_ctx, this patch is to do the
proper cleanup and rollback to the original states for shadowed indirect
context before the workload is abandoned.
v2:
- split the mixed several error paths for better review. (Zhenyu)
v3:
- no return check for clean up functions. (Changbin)
v4:
- expose and reuse the existing release_shadow_wa_ctx. (Zhenyu)
v5:
- move the release function to scheduler.c file. (Zhenyu)
v6:
- move error handling code of intel_gvt_scan_and_shadow_workload
to here. (Zhenyu)
Signed-off-by: fred gao <fred.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
Currently i915 request structure and shadow ring buffer are allocated
before command scan, so it will have to restore to previous states once
any error happens afterwards in the long dispatch_workload path.
This patch is to introduce a reserved ring buffer created at the beginning
of vGPU initialization. Workload will be coped to this reserved buffer and
be scanned first, the i915 request and shadow ring buffer are only
allocated after the result of scan is successful.
To balance the memory usage and buffer alloc time, the coming bigger ring
buffer will be reallocated and kept until more bigger buffer is coming.
v2:
- use kmalloc for the smaller ring buffer, realloc if required. (Zhenyu)
v3:
- remove the dynamically allocated ring buffer. (Zhenyu)
v4:
- code style polish.
- kfree previous allocated buffer once kmalloc failed. (Zhenyu)
Signed-off-by: fred gao <fred.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
For vfio-pci, if the region support MMAP then it should support both
mmap and normal file access. The user-space is free to choose which is
being used. For qemu, we just need add 'x-no-mmap=on' for vfio-pci
option.
Currently GVTg only support MMAP for BAR2. So GVTg will not work when
user turn on x-no-mmap option.
This patch added file style access for BAR2, aka the GPU aperture. We
map the entire aperture partition of active vGPU to kernel space when
guest driver try to enable PCI Memory Space. Then we redirect the file
RW operation from kvmgt to this mapped area.
Link: https://bugzilla.redhat.com/show_bug.cgi?id=1458032
Signed-off-by: Changbin Du <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
For PCI, 64bit bar consumes two BAR registers, but this doesn't mean
both of two BAR are valid. Actually the second BAR is regarded as
reserved in this case. So we shouldn't emulate the second BAR.
Signed-off-by: Changbin Du <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
|
|
Remove our use of 'gperf' for generating perfect hashes from some of our
build tools.
This removal was prompted by Masahiro Yamada sending out a patch that
removes all our pre-generated files, and when I tested it, I noticed
that the gperf version I have (3.1) apparently generates code that no
longer works with out code-base because the function interfaces
generated by gperf have changed.
We really don't care that much, and the gperf people changed their
interfaces in ways that makes it annoying to work with them. Tools that
make it hard to use them should not be used, and the kernel is not at
all interested in some autoconf mess. So remove the gperf dependency
entirely.
It turns out that if you ignore the pre-generated files, the use of
gperf apparently saved us a whopping fifteen lines of code. It
obviously wasn't worth it, considering that the pre-generated files are
about 500 lines.
I sent this out as a patch about three weeks ago, and got absolutely
zero responses. So let's see if anybody notices now that I merge it.
Because there might be serious bugs here, but it WorksForMe(tm).
* gperf-removal:
Remove gperf usage from toolchain
|
|
Pull SCSI updates from James Bottomley:
"This is mostly updates of the usual suspects: lpfc, qla2xxx, hisi_sas,
megaraid_sas, zfcp and a host of minor updates.
The major driver change here is the elimination of the block based
cciss driver in favour of the SCSI based hpsa driver (which now drives
all the legacy cases cciss used to be required for). Plus a reset
handler clean up and the redo of the SAS SMP handler to use bsg lib"
* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (279 commits)
scsi: scsi-mq: Always unprepare before requeuing a request
scsi: Show .retries and .jiffies_at_alloc in debugfs
scsi: Improve requeuing behavior
scsi: Call scsi_initialize_rq() for filesystem requests
scsi: qla2xxx: Reset the logo flag, after target re-login.
scsi: qla2xxx: Fix slow mem alloc behind lock
scsi: qla2xxx: Clear fc4f_nvme flag
scsi: qla2xxx: add missing includes for qla_isr
scsi: qla2xxx: Fix an integer overflow in sysfs code
scsi: aacraid: report -ENOMEM to upper layer from aac_convert_sgraw2()
scsi: aacraid: get rid of one level of indentation
scsi: aacraid: fix indentation errors
scsi: storvsc: fix memory leak on ring buffer busy
scsi: scsi_transport_sas: switch to bsg-lib for SMP passthrough
scsi: smartpqi: remove the smp_handler stub
scsi: hpsa: remove the smp_handler stub
scsi: bsg-lib: pass the release callback through bsg_setup_queue
scsi: Rework handling of scsi_device.vpd_pg8[03]
scsi: Rework the code for caching Vital Product Data (VPD)
scsi: rcu: Introduce rcu_swap_protected()
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/pmladek/printk
Pull printk updates from Petr Mladek:
- Do not allow use of freed init data and code even when boot consoles
are forced to stay. Also check for the init memory more precisely.
- Some code clean up by starting contributors.
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/pmladek/printk:
printk: Clean up do_syslog() error handling
printk/console: Enhance the check for consoles using init memory
printk/console: Always disable boot consoles that use init memory before it is freed
printk: Modify operators of printed_len and text_len
|
|
Thread A: Thread B:
-f2fs_getxattr
-lookup_all_xattrs
-xnid = F2FS_I(inode)->i_xattr_nid;
-f2fs_setxattr
-__f2fs_setxattr
-write_all_xattrs
-truncate_xattr_node
... ...
-write_checkpoint
... ...
-alloc_nid <- nid reuse
-get_node_page
-f2fs_bug_on <- nid != node_footer->nid
It's need a rw_sem to avoid the race
Signed-off-by: Yunlei He <heyunlei@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
This patch refactors get_lock_data_page() to handle encryption case directly.
In order to do that, it introduces common f2fs_submit_page_read().
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
