Commit graph

932,869 commits

Author SHA1 Message Date
Christian König
3a80e92b6e drm/amdgpu: split finding idle VMID into separate function
No functional change, but makes it easier to maintain the code.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:14 -05:00
Christian König
8fe27f8fa0 drm/amdgpu: make VMID assignment more fair v2
Similar to finding an idle one before reuse.

This guarantees fairness between processes. Otherwise process with a reserved
VMID have an unfair advantage while scheduling jobs.

v2: improve commit message

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:13 -05:00
Bas Nieuwenhuizen
a20ee0b1f8 drm/amdgpu: Fix always_valid bos multiple LRU insertions.
If these bos are evicted and are in the validated list
things blow up, so do not put them in there. Notably,
that tries to add the bo to the LRU twice, which results
in a BUG_ON in ttm_bo.c.

While for the bo_list an alternative would be to not allow
always valid bos in there, that does not work for the user
fence.

v2: Fixed whitespace issue pointed out by checkpatch.pl

Signed-off-by: Bas Nieuwenhuizen <basni@chromium.org>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
2018-02-19 14:19:13 -05:00
David S. Miller
cc35b39678 Merge branch 'pernet_ops-conversions-part-2'
Kirill Tkhai says:

====================
Converting pernet_operations (part #2)

This patchset continues to review and to convert pernet_operations
to async. There are mostly ipv6, also some regular used netfilter
pernet_operations are involved. One more converted is cfg80211_pernet_ops.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:12 -05:00
Andrey Grodzovsky
062d2e6af7 Revert "drm/amdgpu/gfx8: Fix compute ring failure after resetting"
This reverts commit 75737cb4eb.

Fixes compute rings test failure on bare metal during full GPU reset.

RCA:
the ring buffer has to be filled with valid packets (such as NOPs) first
before submitting MAP_QUEUEs packet into KIQ. Once a compute engine is mapped,
it will immediately execute the ring buffer if the RTPR is not equal to the
WTPR from the MQD. It could lead to engine hang if the ring buffer filled
with random data.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:12 -05:00
Kirill Tkhai
da349fad80 net: Convert iptable_filter_net_ops
These pernet_operations register and unregister
net::ipv4.iptable_filter table. Since there are
no packets in-flight at the time of exit method
is working, iptables rules should not be touched.
Also, pernet_operations should not send ipv4
packets each other. So, it's safe to mark them
async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:12 -05:00
Christian König
f732b6b3c0 drm/amdgpu: move waiting for VM flush into gmc_v9_0_emit_flush_gpu_tlb
Keep that at a common place instead of spread over all engines.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:12 -05:00
Kirill Tkhai
4d6b80762b net: Convert ip_tables_net_ops, udplite6_net_ops and xt_net_ops
ip_tables_net_ops and udplite6_net_ops create and destroy /proc entries.
xt_net_ops does nothing.

So, we are able to mark them async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:12 -05:00
Christian König
2b124b0b70 drm/amdgpu: implement vcn_v1_0_(dec|enc)_ring_emit_reg_wait v2
Add emit_reg_wait implementation for VCN v1.

v2: cleanup the existing code as well

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:11 -05:00
Kirill Tkhai
5fc094f5b8 net: Convert ip6_frags_ops
Exit methods calls inet_frags_exit_net() with global ip6_frags
as argument. So, after we make the pernet_operations async,
a pair of exit methods may be called to iterate this hash table.
Since there is inet_frag_worker(), which already may work
in parallel with inet_frags_exit_net(), and it can make the same
cleanup, that inet_frags_exit_net() does, it's safe. So we may
mark these pernet_operations as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:11 -05:00
Kirill Tkhai
d16784d9fb net: Convert fib6_net_ops, ipv6_addr_label_ops and ip6_segments_ops
These pernet_operations register and unregister tables
and lists for packets forwarding. All of the entities
are per-net. Init methods makes simple initializations,
and since net is not visible for foreigners at the time
it is working, it can't race with anything. Exit method
is executed when there are only local devices, and there
mustn't be packets in-flight. Also, it looks like no one
pernet_operations want to send ipv6 packets to foreign
net. The same reasons are for ipv6_addr_label_ops and
ip6_segments_ops. So, we are able to mark all them as
async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:11 -05:00
Kirill Tkhai
b489141369 net: Convert xfrm6_net_ops
These pernet_operations create sysctl tables and
initialize net::xfrm.xfrm6_dst_ops used for routing.
It doesn't look like another pernet_operations send
ipv6 packets to foreign net namespaces, so it should
be safe to mark the pernet_operations as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:11 -05:00
Kirill Tkhai
a7852a76f4 net: Convert ip6_flowlabel_net_ops
These pernet_operations create and destroy /proc entries.
ip6_fl_purge() makes almost the same actions as timer
ip6_fl_gc_timer does, and as it can be executed in parallel
with ip6_fl_purge(), two parallel ip6_fl_purge() may be
executed. So, we can mark it async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:11 -05:00
Kirill Tkhai
ac34cb6c0c net: Convert ping_v6_net_ops
These pernet_operations only register and unregister /proc
entries, so it's possible to mark them async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:11 -05:00
Kirill Tkhai
58708caef5 net: Convert ipv6_sysctl_net_ops
These pernet_operations create and destroy sysctl tables.
They are not touched by another net pernet_operations.
So, it's possible to execute them in parallel with others.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:11 -05:00
Christian König
d949315a66 drm/amdgpu: implement vce_v4_0_emit_reg_wait v2
Add emit_reg_wait implementation for VCE v4.

v2: call new function directly from existing code

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:10 -05:00
Kirill Tkhai
fef65a2c6c net: Convert tcpv6_net_ops
These pernet_operations create and destroy net::ipv6.tcp_sk
socket, which is used in tcp_v6_send_response() only. It looks
like foreign pernet_operations don't want to set ipv6 connection
inside destroyed net, so this socket may be created in destroyed
in parallel with anything else. inet_twsk_purge() is also safe
for that, as described in patch for tcp_sk_ops. So, it's possible
to mark them as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:10 -05:00
Christian König
38d32a7564 drm/amdgpu: implement uvd_v7_0_(enc_|)ring_emit_reg_wait v2
Add emit_reg_wait implementation for UVD v7.

v2: call new function directly from the existing code

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:10 -05:00
Kirill Tkhai
7b7dd180b8 net: Convert fib6_rules_net_ops
These pernet_operations register and unregister
net::ipv6.fib6_rules_ops, which are used for
routing. It looks like there are no pernet_operations,
which send ipv6 packages to another net, so we
are able to mark them as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:10 -05:00
Kirill Tkhai
85ca51b2a2 net: Convert ipv6_inetpeer_ops
net->ipv6.peers is dereferenced in three places via inet_getpeer_v6(),
and it's used to handle skb. All the users of inet_getpeer_v6() do not
look like be able to be called from foreign net pernet_operations, so
we may mark them as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:10 -05:00
Kirill Tkhai
509114112d net: Convert raw6_net_ops, udplite6_net_ops, ipv6_proc_ops, if6_proc_net_ops and ip6_route_net_late_ops
These pernet_operations create and destroy /proc entries
and safely may be converted and safely may be mark as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:10 -05:00
Kirill Tkhai
1a2e93329d net: Convert icmpv6_sk_ops, ndisc_net_ops and igmp6_net_ops
These pernet_operations create and destroy net::ipv6.icmp_sk
socket, used to send ICMP or error reply.

Nobody can dereference the socket to handle a packet before
net is initialized, as there is no routing; nobody can do
that in parallel with exit, as all of devices are moved
to init_net or destroyed and there are no packets it-flight.
So, it's possible to mark these pernet_operations as async.

The same for ndisc_net_ops and for igmp6_net_ops. The last
one also creates and destroys /proc entries.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:10 -05:00
Kirill Tkhai
b01a59a488 net: Convert ip6mr_net_ops
These pernet_operations create and destroy /proc entries,
populate and depopulate net::rules_ops and multiroute table.
All the structures are pernet, and they are not touched
by foreign net pernet_operations. So, it's possible to mark
them async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:10 -05:00
Christian König
5ad68ee4ca drm/amdgpu: implement sdma_v4_0_ring_emit_reg_wait
Add emit_reg_wait implementation for SDMA v4.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:09 -05:00
Kirill Tkhai
9c537ca155 net: Convert cfg80211_pernet_ops
This patch finishes converting pernet_operations
registered in net/wireless directory.

These pernet_operations have only exit method,
which moves devices to init_net. This action
is not pernet_operations-specific, and function
cfg80211_switch_netns() may be called all time
during the system life. All necessary protection
against concurrent cfg80211_pernet_exit() is made
by rtnl_lock(). So, cfg80211_pernet_ops is able
to be marked as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:09 -05:00
Christian König
230fcc34a8 drm/amdgpu: implement gfx_v9_0_ring_emit_reg_wait
Implement emit_reg_wait for gfx v9.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:09 -05:00
Kirill Tkhai
753d525a08 net: Convert inet6_net_ops
init method initializes sysctl defaults, allocates
percpu arrays and creates /proc entries.
exit method reverts the above.

There are no pernet_operations, which are interested
in the above entities of foreign net namespace, so
inet6_net_ops are able to be marked as async.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2018-02-19 14:19:09 -05:00
Christian König
c1e877da57 drm/amdgpu: add new emit_reg_wait callback
Allows us to wait for a register value/mask on a ring.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:08 -05:00
Christian König
491355937a drm/amdgpu: fix DW estimation on VI
Forgot to update that during recent changes.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:08 -05:00
Rex Zhu
61e208b16e drm/amd/pp: Fix sysfs pp_dpm_pcie bug on CI/VI
when echo "01">pp_dpm_pcie
the pcie dpm will fix in highest link speed.
But user should expect auto speed between
level 0 and level1

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:07 -05:00
Rex Zhu
337ecd6a98 drm/amd/pp: Enable pp_set_power_profile_mode under manual mode
Only user enter manual performance mode, driver allow user
configure the sclk/mclk dpm parameters through sysfs
pp_power_profile_mode.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:06 -05:00
Rex Zhu
948fd66c94 drm/amd/pp: Refine pp_dpm_force_clock_level functions
Only when user set manual performance mode, driver
enable pp_dpm_force_clock_level.
so check the mode in pp_dpm_force_clock_level,
and delete the same logic in callback functions.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:06 -05:00
Rex Zhu
6fa3ef3a9c drm/amd/pp: Delete debug info when set power_profile_mode
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:05 -05:00
Christian König
6b034e25c8 drm/amdgpu: limit the VM address space with older VCE FW versions
Older VCE FW versions are buggy and can't work with 48bit address
spaces.

RFC: Should we limit the address space or just reject loading the older
VCE firmware?

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:05 -05:00
Luis de Bethencourt
e2751493fc drm: Fix trailing semicolon
The trailing semicolon is an empty statement that does no operation.
Removing it since it doesn't do anything.

Signed-off-by: Luis de Bethencourt <luisbg@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:04 -05:00
Luis de Bethencourt
122fe39da4 drm: amd: Fix trailing semicolons
The trailing semicolon is an empty statement that does no operation.
Removing the two instances of them since they don't do anything.

Signed-off-by: Luis de Bethencourt <luisbg@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:03 -05:00
Alex Deucher
3fa203af4b drm/amdgpu: set DRIVER_ATOMIC flag early
The atomic debugfs stuff gets created in drm_dev_alloc()
but this gets called before we've enumerated all of our
IPs, so move the DRIVER_ATOMIC flag setting to fix that.

Since DRIVER_ATOMIC is a driver flag it's currently global
to the driver so setting it affects all GPUs driven by the
driver.  Unfortunately, not all GPUs support atomic.  Warn
the user if that is the case.

This is the same as our current behavior, but at least the
atomic debugfs stuff gets created now.

Acked-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:03 -05:00
Alex Deucher
367e66870e drm/amdgpu: remove DC special casing for KB/ML
It seems to be working now.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=102372
Reviewed-by: Mike Lothian <mike@fireburn.co.uk>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:02 -05:00
Tom St Denis
7ca34ddc78 drm/ttm: Simplify ttm_dma_page_put()
Remove redundant store of return code.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:02 -05:00
Tom St Denis
42bdbb6e0f drm/ttm: Fix coding style in ttm_dma_pool_alloc_new_pages()
Add missing {} braces.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:01 -05:00
Tom St Denis
ddde985c41 drm/ttm: Fix coding style in ttm_tt_swapout()
Add missing {} braces.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:01 -05:00
Tom St Denis
07d48da4f4 drm/ttm: Simplify ttm_eu_reserve_buffers()
Hoist the comparison of the ret to -EDEADLK above
the two code paths to simplify the function.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:19:00 -05:00
Tom St Denis
de8dfb8e34 drm/ttm: Remove unncessary retval from ttm_bo_vm_fault()
The dual ret/retval was more complex than need be.  Now
we drop the retval variable and assign the appropriate VM
codes to ret instead.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:59 -05:00
Tom St Denis
449f797a94 drm/ttm: Fix coding style in ttm_bo_move_memcpy()
Add missing {} braces.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:59 -05:00
Tom St Denis
add3d95d73 drm/ttm: Simplify ttm_dma_find_pool() (v2)
Flip the logic of the comparison and remove
the redudant variable for the pool address.

(v2): Remove {} bracing.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:58 -05:00
Tom St Denis
c68edaa0c1 drm/ttm: Fix coding style in ttm_pool_store()
Correct missing {} style.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:58 -05:00
Tom St Denis
5b4262d7a2 drm/ttm: Change ttm_tt page allocations to return errors
Explicitly return errors in ttm_tt_alloc_page_directory() and
ttm_dma_tt_alloc_page_directory() instead of relying on
further logic to detect errors.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:57 -05:00
Tom St Denis
420457acfb drm/ttm: Add a default BO destructor to simplify code (v2)
(v2): Remove stray ; noticed by Felix

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:56 -05:00
Tom St Denis
43c7c41b25 drm/ttm: Fix coding style in ttm_bo.c
Correct indentation and {} brace style.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:56 -05:00
Christian König
4584312d38 drm/amdgpu: fill only the lower range with ATS entries v2
At least on x86-64 the upper range is purely used by the kernel,
avoid creating any ATS mappings there as security precaution and to
allow proper page fault reporting in the upper range.

v2: remove unused variable

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2018-02-19 14:18:55 -05:00