Since RK3288 DMAC's burst length only support max to 4, here
set maxburst of playback and capture dma data to 4.
Signed-off-by: Jianqun Xu <jay.xu@rock-chips.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
According to description about "Transmit Data Level",
This bit field controls the level at which a DMA request
is made by the transmit logic.
It is equal to the watermark level.
That is, the dma_tx_req signal is generated when the number
of valid data entries in the TXFIFO
(TXFIFO0 if CSR=00
TXFIFO1 if CSR=01
TXFIFO2 if CSR=10
TXFIFO3 if CSR=11)
is equal to or below this field value.
Different to receive data level, transmit data level does not need
to "-1".
Signed-off-by: Jianqun Xu <jay.xu@rock-chips.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
A 1-byte burst size is rather inefficient and has been shown to cause
TX issues during testing. Increase the DMA burst size to 4-bytes for
both RX and TX DMA when using the 8-bit FIFO.
Signed-off-by: Andrew Bresticker <abrestic@chromium.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
For block span more than 1 section, when allocate it from
a free block, we need allocate the remain buffers within
the block, and then continue alloc the rest of needed
size buffer.
Here also make sure this free block is moved from free
list to used list, and add it to block_list which may
be used for power gating disabling later.
Signed-off-by: Jie Yang <yang.jie@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
snd_kcontrol_chip should return snd_soc_component instead of
snd_soc_codec
Signed-off-by: Fang, Yang A <yang.a.fang@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
skb_scrub_packet() is called when a packet switches between a context
such as between underlay and overlay, between namespaces, or between
L3 subnets.
While we already scrub the packet mark, connection tracking entry,
and cached destination, the security mark/context is left intact.
It seems wrong to inherit the security context of a packet when going
from overlay to underlay or across forwarding paths.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Acked-by: Flavio Leitner <fbl@sysclose.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When vlan tags are stacked, it is very likely that the outer tag is stored
in skb->vlan_tci and skb->protocol shows the inner tag's vlan_proto.
Currently netif_skb_features() first looks at skb->protocol even if there
is the outer tag in vlan_tci, thus it incorrectly retrieves the protocol
encapsulated by the inner vlan instead of the inner vlan protocol.
This allows GSO packets to be passed to HW and they end up being
corrupted.
Fixes: 58e998c6d2 ("offloading: Force software GSO for multiple vlan tags.")
Signed-off-by: Toshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pravin B Shelar says:
====================
openvswitch: datapath fixes
Following patch series is mostly targeted to MPLS fixes. other
patches are related datapth transmit path error handling.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
In case of error vxlan_xmit_one() can free already freed skb.
Also fixes memory leak of dst-entry.
Fixes: acbf74a763 ("vxlan: Refactor vxlan driver to make use
of the common UDP tunnel functions").
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Today vport-send has complex error handling because it involves
freeing skb and updating stats depending on return value from
vport send implementation.
This can be simplified by delegating responsibility of freeing
skb to the vport implementation for all cases. So that
vport-send needs just update stats.
Fixes: 91b7514cdf ("openvswitch: Unify vport error stats
handling")
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
MPLS GSO needs to know inner most protocol to process GSO packets.
Fixes: 25cd9ba0ab ("openvswitch: Add basic MPLS support to
kernel").
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Linux stack does not implement GSO for packet with multiple
encapsulations. Therefore there was check in MPLS action
validation to detect such case, But this check introduced
bug which deleted one or more actions from actions list.
Following patch removes this check to fix the validation.
Fixes: 25cd9ba0ab ("openvswitch: Add basic MPLS support to
kernel").
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Reported-by: Srinivas Neginhal <sneginha@vmware.com>
Acked-by: Jarno Rajahalme <jrajahalme@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
MPLS and Tunnel GSO does not work together. Reject packet which
request such GSO.
Fixes: 0d89d2035f ("MPLS: Add limited GSO support").
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes MPLS GSO for case when mpls is compiled as kernel module.
Fixes: 0d89d2035f ("MPLS: Add limited GSO support").
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch rearranges the loop in net_rx_action to reduce the
amount of jumping back and forth when reading the code.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
We should only perform the softnet_break check after we have polled
at least one device in net_rx_action. Otherwise a zero or negative
setting of netdev_budget can lock up the whole system.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
The commit d75b1ade56 (net: less
interrupt masking in NAPI) required drivers to leave poll_list
empty if the entire budget is consumed.
We have already had two broken drivers so let's add a check for
this.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch creates a new function napi_poll and moves the napi
polling code from net_rx_action into it.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Gateway having bandwidth_down equal to zero are not accepted
at all and so never added to the Gateway list.
For this reason checking the bandwidth_down member in
batadv_gw_out_of_range() is useless.
This is probably a copy/paste error and this check was supposed
to be "!gw_node" only. Moreover, the way the check is written
now may also lead to a NULL dereference.
Fix this by rewriting the if-condition properly.
Introduced by 414254e342
("batman-adv: tvlv - gateway download/upload bandwidth container")
Signed-off-by: Antonio Quartulli <antonio@meshcoding.com>
Reported-by: David Binderman <dcb314@hotmail.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
The fragmentation code was replaced in 610bfc6bc9
("batman-adv: Receive fragmented packets and merge") by an implementation which
can handle up to 16 fragments of a packet. The packet is prepared for the split
in fragments by the function batadv_frag_send_packet and the actual split is
done by batadv_frag_create.
Both functions calculate the size of a fragment themself. But their calculation
differs because batadv_frag_send_packet also subtracts ETH_HLEN. Therefore,
the check in batadv_frag_send_packet "can a full fragment can be created?" may
return true even when batadv_frag_create cannot create a full fragment.
The function batadv_frag_create doesn't check the size of the skb before
splitting it and therefore might try to create a larger fragment than the
remaining buffer. This creates an integer underflow and an invalid len is given
to skb_split.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The fragmentation code was replaced in 610bfc6bc9
("batman-adv: Receive fragmented packets and merge"). The new code provided a
mostly unused parameter skb for the merging function. It is used inside the
function to calculate the additionally needed skb tailroom. But instead of
increasing its own tailroom, it is only increasing the tailroom of the first
queued skb. This is not correct in some situations because the first queued
entry can be a different one than the parameter.
An observed problem was:
1. packet with size 104, total_size 1464, fragno 1 was received
- packet is queued
2. packet with size 1400, total_size 1464, fragno 0 was received
- packet is queued at the end of the list
3. enough data was received and can be given to the merge function
(1464 == (1400 - 20) + (104 - 20))
- merge functions gets 1400 byte large packet as skb argument
4. merge function gets first entry in queue (104 byte)
- stored as skb_out
5. merge function calculates the required extra tail as total_size - skb->len
- pskb_expand_head tail of skb_out with 64 bytes
6. merge function tries to squeeze the extra 1380 bytes from the second queued
skb (1400 byte aka skb parameter) in the 64 extra tail bytes of skb_out
Instead calculate the extra required tail bytes for skb_out also using skb_out
instead of using the parameter skb. The skb parameter is only used to get the
total_size from the last received packet. This is also the total_size used to
decide that all fragments were received.
Reported-by: Philipp Psurek <philipp.psurek@gmail.com>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Acked-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit cecda693a9 ("net: keep original skb
which only needs header checking during software GSO") keeps the original
skb for packets that only needs header check, but it doesn't drop the
packet if software segmentation or header check were failed.
Fixes cecda693a9 ("net: keep original skb which only needs header checking during software GSO")
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit 355a701838.
This had some bad side effects under normal operation, and should
have been dropped earlier.
Signed-off-by: Dave Airlie <airlied@redhat.com>
- Display MEC fw version in topology. Without this, the HSA userspace
stack is broken.
- Init apertures information only once per process
* tag 'amdkfd-fixes-2014-12-23' of git://people.freedesktop.org/~gabbayo/linux:
amdkfd: init aperture once per process
amdkfd: Display MEC fw version in topology node
drm/radeon: Add implementation of get_fw_version
drm/amd: Add get_fw_version to kfd-->kgd interface
The Int340x thermal provides a processor thermal device, which
is used to control processor thermal states. These devices are
either reported as a PCI device or an ACPI device. This
device provides power limits, control states and optional
temperature.
This change implements minimal requirements to expose processor
power limits which can be used during thermal power limiting.
Power limits are exposed via an attribute group called
"power_limits" under the device. The exported attributes
are:
power_limit_0_max_uw
power_limit_1_max_uw
power_limit_0_min_uw
power_limit_1_min_uw
power_limit_0_tmin_us
power_limit_1_tmin_us
power_limit_0_tmax_us
power_limit_1_tmax_us
power_limit_0_step_uw
power_limit_1_step_uw
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Pull audit fixes from Paul Moore:
"Four patches to fix various problems with the audit subsystem, all are
fairly small and straightforward.
One patch fixes a problem where we weren't using the correct gfp
allocation flags (GFP_KERNEL regardless of context, oops), one patch
fixes a problem with old userspace tools (this was broken for a
while), one patch fixes a problem where we weren't recording pathnames
correctly, and one fixes a problem with PID based filters.
In general I don't think there is anything controversial with this
patchset, and it fixes some rather unfortunate bugs; the allocation
flag one can be particularly scary looking for users"
* 'upstream' of git://git.infradead.org/users/pcmoore/audit:
audit: restore AUDIT_LOGINUID unset ABI
audit: correctly record file names with different path name types
audit: use supplied gfp_mask from audit_buffer in kauditd_send_multicast_skb
audit: don't attempt to lookup PIDs when changing PID filtering audit rules
First set of fixes for current -rc cycle. There are
a couple of build break fixes after Tony Lindgren's
recent MUSB patchset, some memory leak fixes also
with MUSB, a use-after-free fix with the UAC1
function. Atmel UDC got a fix for a possible hang
and another for DMA setting, while dwc2 learned to
kill requests in ->udc_stop() which fixes a few leaks
too.
One new device support here, dwc3 now supports Intel's
Sunrise Point.
Signed-off-by: Felipe Balbi <balbi@ti.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=9r9T
-----END PGP SIGNATURE-----
Merge tag 'fixes-for-v3.19-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-linus
Felipe writes:
usb: fixes for v3.19-rc2
First set of fixes for current -rc cycle. There are
a couple of build break fixes after Tony Lindgren's
recent MUSB patchset, some memory leak fixes also
with MUSB, a use-after-free fix with the UAC1
function. Atmel UDC got a fix for a possible hang
and another for DMA setting, while dwc2 learned to
kill requests in ->udc_stop() which fixes a few leaks
too.
One new device support here, dwc3 now supports Intel's
Sunrise Point.
Signed-off-by: Felipe Balbi <balbi@ti.com>
A regression was caused by commit 780a7654cee8:
audit: Make testing for a valid loginuid explicit.
(which in turn attempted to fix a regression caused by e1760bd)
When audit_krule_to_data() fills in the rules to get a listing, there was a
missing clause to convert back from AUDIT_LOGINUID_SET to AUDIT_LOGINUID.
This broke userspace by not returning the same information that was sent and
expected.
The rule:
auditctl -a exit,never -F auid=-1
gives:
auditctl -l
LIST_RULES: exit,never f24=0 syscall=all
when it should give:
LIST_RULES: exit,never auid=-1 (0xffffffff) syscall=all
Tag it so that it is reported the same way it was set. Create a new
private flags audit_krule field (pflags) to store it that won't interact with
the public one from the API.
Cc: stable@vger.kernel.org # v3.10-rc1+
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: Paul Moore <pmoore@redhat.com>
In Linux 3.18 and below, GCC hoists the lsl instructions in the
pvclock code all the way to the beginning of __vdso_clock_gettime,
slowing the non-paravirt case significantly. For unknown reasons,
presumably related to the removal of a branch, the performance issue
is gone as of
e76b027e64 x86,vdso: Use LSL unconditionally for vgetcpu
but I don't trust GCC enough to expect the problem to stay fixed.
There should be no correctness issue, because the __getcpu calls in
__vdso_vlock_gettime were never necessary in the first place.
Note to stable maintainers: In 3.18 and below, depending on
configuration, gcc 4.9.2 generates code like this:
9c3: 44 0f 03 e8 lsl %ax,%r13d
9c7: 45 89 eb mov %r13d,%r11d
9ca: 0f 03 d8 lsl %ax,%ebx
This patch won't apply as is to any released kernel, but I'll send a
trivial backported version if needed.
Fixes: 51c19b4f59 x86: vdso: pvclock gettime support
Cc: stable@vger.kernel.org # 3.8+
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Commit 1290a958d4 ("usb: phy: propagate __of_usb_find_phy()'s error on
failure") broke platforms that rely on deferred probing to order probing
of PHY and host controller drivers. The reason is that the commit simply
propagates errors from __of_usb_find_phy(), which returns -ENODEV if no
PHY has been registered yet for a given device tree node. The only case
in which -EPROBE_DEFER would now be returned is if try_module_get() did
fail, which does not make sense.
The correct thing to do is to return -EPROBE_DEFER if a PHY hasn't been
registered yet. The only condition under which it makes sense to return
-ENODEV is if the device tree node representing the PHY has been
disabled (via the status property) because in that case the PHY will
never be registered.
This patch addresses the problem by making __of_usb_find_phy() return an
appropriate error code while keeping in line with the above-mentioned
commit to propagate error codes rather than overwriting them. At the
same time the check for a valid PHY is decoupled from the check for the
try_module_get() call and a separate error code is returned if the
latter fails.
Fixes: 1290a95 (usb: phy: propagate __of_usb_find_phy()'s error on failure)
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch adds pgd_page definition in order to keep supporting
HAVE_GENERIC_RCU_GUP configuration. In addition, it changes pud_page
expression to align with pmd_page for readability.
An introduction of pgd_page resolves the following build breakage
under 4KB + 4Level memory management combo.
mm/gup.c: In function 'gup_huge_pgd':
mm/gup.c:889:2: error: implicit declaration of function 'pgd_page' [-Werror=implicit-function-declaration]
head = pgd_page(orig);
^
mm/gup.c:889:7: warning: assignment makes pointer from integer without a cast
head = pgd_page(orig);
Cc: Will Deacon <will.deacon@arm.com>
Cc: Steve Capper <steve.capper@linaro.org>
Signed-off-by: Jungseok Lee <jungseoklee85@gmail.com>
[catalin.marinas@arm.com: remove duplicate pmd_page definition]
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The usual defconfig tweaks, this time:
- FHANDLE and AUTOFS4_FS to keep systemd happy
- PID_NS, QUOTA and KEYS to keep LTP happy
- Disable DEBUG_PREEMPT, as this *really* hurts performance
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
On arm64 the TTBR0_EL1 register is set to either the reserved TTBR0
page tables on boot or to the active_mm mappings belonging to user space
processes, it must never be set to swapper_pg_dir page tables mappings.
When a CPU is booted its active_mm is set to init_mm even though its
TTBR0_EL1 points at the reserved TTBR0 page mappings. This implies
that when __cpu_suspend is triggered the active_mm can point at
init_mm even if the current TTBR0_EL1 register contains the reserved
TTBR0_EL1 mappings.
Therefore, the mm save and restore executed in __cpu_suspend might
turn out to be erroneous in that, if the current->active_mm corresponds
to init_mm, on resume from low power it ends up restoring in the
TTBR0_EL1 the init_mm mappings that are global and can cause speculation
of TLB entries which end up being propagated to user space.
This patch fixes the issue by checking the active_mm pointer before
restoring the TTBR0 mappings. If the current active_mm == &init_mm,
the code sets the TTBR0_EL1 to the reserved TTBR0 mapping instead of
switching back to the active_mm, which is the expected behaviour
corresponding to the TTBR0_EL1 settings when __cpu_suspend was entered.
Fixes: 95322526ef ("arm64: kernel: cpu_{suspend/resume} implementation")
Cc: <stable@vger.kernel.org> # 3.14+: 18ab7db
Cc: <stable@vger.kernel.org> # 3.14+: 714f599
Cc: <stable@vger.kernel.org> # 3.14+: c3684fb
Cc: <stable@vger.kernel.org> # 3.14+
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The perf report --children can be called with callchain disabled so no
need to append callchains. Actually the root of callchain tree is not
initialized properly in this case.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1419223455-4362-7-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The output will look like below. (I added an error into ui__init() for
the test).
$ perf report
perf: Segmentation fault
-------- backtrace --------
perf[0x503781]
/usr/lib/libc.so.6(+0x33b20)[0x7f1a14f04b20]
perf(ui__init+0xd5)[0x503645]
perf(setup_browser+0x97)[0x4ce4e7]
perf(cmd_report+0xcea)[0x4392ba]
perf[0x428493]
perf(main+0x60a)[0x427c0a]
/usr/lib/libc.so.6(__libc_start_main+0xf0)[0x7f1a14ef1040]
perf[0x427d29]
[0x0]
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1419223455-4362-4-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Sometimes it takes a long time to resort hist entries for output in case
of a large data file. Show a progress bar window and inform user.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/1419223455-4362-3-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Commit 9def39be4e ("x86: Support compiling out human-friendly
processor feature names") made two source file targets
conditional. Such conditional targets will not be cleaned
automatically by make mrproper.
Fix by adding explicit clean-files targets for the two files.
Fixes: 9def39be4e ("x86: Support compiling out human-friendly processor feature names")
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Cc: Josh Triplett <josh@joshtriplett.org>
Link: http://lkml.kernel.org/r/1419335863-10608-1-git-send-email-bjorn@mork.no
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Make sure this fetches 16-bits port data from the register.
Remove casting to make sparse happy, not needed anymore.
Signed-off-by: leroy christophe <christophe.leroy@c-s.fr>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
When allocating space for load_balance_mask, in sched_init, when
CPUMASK_OFFSTACK is set, we've managed to spill over
KMALLOC_MAX_SIZE on our 6144 core machine. The patch below
breaks up the allocations so that they don't overflow the max
alloc size. It also allocates the masks on the the node from
which they'll most commonly be accessed, to minimize remote
accesses on NUMA machines.
Suggested-by: George Beshers <gbeshers@sgi.com>
Signed-off-by: Alex Thorlton <athorlton@sgi.com>
Cc: George Beshers <gbeshers@sgi.com>
Cc: Russ Anderson <rja@sgi.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1418928270-148543-1-git-send-email-athorlton@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The old scheme can lead to failure in certain cases - the
problem is that after bumping step_size the next (non-final)
iteration is only guaranteed to make available a memory block
the size of what step_size was before. E.g. for a memory block
[0,3004600000) we'd have:
iter start end step amount
1 3004400000 30045fffff 2M 2M
2 3004000000 30043fffff 64M 4M
3 3000000000 3003ffffff 2G 64M
4 2000000000 2fffffffff 64G 64G
Yet to map 64G with 4k pages (as happens e.g. under PV Xen) we
need slightly over 128M, but the first three iterations made
only about 70M available.
The condition (new_mapped_ram_size > mapped_ram_size) for
bumping step_size is just not suitable. Instead we want to bump
it when we know we have enough memory available to cover a block
of the new step_size. And rather than making that condition more
complicated than needed, simply adjust step_size by the largest
possible factor we know we can cover at that point - which is
shifting it left by one less than the difference between page
table level shifts. (Interestingly the original STEP_SIZE_SHIFT
definition had a comment hinting at that having been the
intention, just that it should have been PUD_SHIFT-PMD_SHIFT-1
instead of (PUD_SHIFT-PMD_SHIFT)/2, and of course for non-PAE
32-bit we can't really use these two constants as they're equal
there.)
Furthermore the comment in get_new_step_size() didn't get
updated when the bottom-down mapping logic got added. Yet while
an overflow (flushing step_size to zero) of the shift doesn't
matter for the top-down method, it does for bottom-up because
round_up(x, 0) = 0, and an upper range boundary of zero can't
really work well.
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/54945C1E020000780005114E@mail.emea.novell.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
There is no reason for having it and, with commit 250a1ac685 ("x86,
smpboot: Remove pointless preempt_disable() in
native_smp_prepare_cpus()"), it prevents HVM guests from booting.
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Mark Brown