Userspace sends tcp connection (sock) destroy on network permission
change. Kernel though doesn't send reset for the connections in
SYN-SENT state and these connections continue to remain. Even as
per RFC 793, there is no hard rule to not send RST on ABORT in
this state. Modify tcp_abort and tcp_disconnect behavior to send
RST for connections in syn-sent state to avoid lingering
connections on network switch.
CRs-Fixed: 2396964
Change-Id: If931fb979e519c474697d82d8c9dbaf0edb626e3
Signed-off-by: Devi Sandeep Endluri V V <dendluri@codeaurora.org>
If we query the OPP table when it doesn't exist we get an error. If
we populate it again with the same operating points we get an error.
So just skip the table create if it already exists and go with
whatever we have.
Change-Id: Ic0dedbad3d5bd24b35867a82873a6ece555d8314
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Make the fallback path for claiming platform devices trigger only if a
new module parameter is specified:
serdev_ttyport.pdev_tty_port=ttyS2
Bug: 146517987
Change-Id: Ibf331ad6e6d8712a405921530f217f7122428b13
Signed-off-by: Alistair Delva <adelva@google.com>
Git-commit: 3a97aabe4f
Git-repo: https://android.googlesource.com/kernel/common/
Signed-off-by: Akash Asthana <akashast@codeaurora.org>
This change adds the missing set_mode callback for RUMI UFS PHY driver.
Change-Id: Ib67ebf63690a1d01172d4e9c9a0c253a8595f5fb
Signed-off-by: Can Guo <cang@codeaurora.org>
Add IPC verbose log to print OOB and DB mode event and
the door bell ring status. Include the DB ring status
in data channel queue API. Add OOB/DB mode counter.
This helps to identify how long MHI host took to ring
the doorbell and how many times these events are getting
triggered. Reset mode_change count in debugfs show API
in order to track this for a given use case.
Change-Id: I2215554bff6d66098a8c7d01f587c1d726443181
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
After getting OOB driver checks for transfer ring empty
condition and rings the channel ring doorbell without
making sure if there are minimum 8 TREs queued for
GSI to perform DMA. Otherwise there is a possibility of
Host ringing doorbell after OOB with same write pointer
which was already processed in polling mode.
Change-Id: Idda28aadbd1af9b5857db71d98237ef4761427bd
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
Remove unused firmware loader worker thread as MHI has moved
on to using pre-existing worker thread in order to serialize
the process.
Change-Id: I71349c97680ec1c0bf17e8f37ef8b046487a2b19
Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
In some cases, the special purpose event scheduler may not run
if the interrupt comes close to mission mode entry point and if
no suspend/resume activity follows. Do a check to see if any
pending events need to be processed.
Change-Id: Ic3bf22562f250ec4bb46c14f58617c3b0d5de883
Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
Some device requested special purpose events can take a long
time to get processed on host. Fix that by moving to dedicated
worker thread.
Change-Id: Ibfe7c56b85cb58b9ee67e41ea1139702d475ffee
Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
Upon power up driver queues firmware worker thread if ee is in PBL.
Firmware worker is blocked with a timeout until state worker gets a
chance to run and unblock firmware worker. In case state worker gets a
chance to run after firmware worker timed out results into endpoint
boot up failure. Fix this issue by directly handling firmware loading
from state worker thread.
Change-Id: Id28c5a3eaa3b8075a7eb0185f7e08f21174e1d0d
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
commit af34613293 ("usb: dwc3-msm: Get usb power_supply from
device tree") changed the power_supply_get_by_name() to
get_by_phandle(), however this does not work if the device
pointed to by the phandle registers multiple power_supply
objects (including battery & wireless as is the case for
QTI PMICs) as we are not guaranteed to get back the USB psy.
Hence revert back to using power_supply_get_by_name() for
the named "usb" psy. We can keep using the "qcom,usb-charger"
DT property but just treat it as boolean instead so that we
can support USB controllers that may not be tied to a charger.
Change-Id: Ic7e4c58b096b0d6210cd55f1ff6edf21ffc5b4b1
Signed-off-by: Jack Pham <jackp@codeaurora.org>
This change is for general scheduler improvements.
Change-Id: Icd64897b7ad5fb8fa5afc418aaca48bdfefca2ef
Signed-off-by: Shaleen Agrawal <shalagra@codeaurora.org>
Enable KERNEL_HEADER_TEST and for basic sanity testing on kernel headers.
Change-Id: Ic0dedbad786a8ee23a236298e79ad592d92b4193
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
The hyp_core_ctl trace file uses a structure that is only defined
internally in the file so it does not pass KERNEL_HEADER_TEST in the
global header location. Since it is only used by this one file it
can be safely moved to the local directory.
Change-Id: Ic0dedbad4285ffcc750839ebc673029193733437
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Add #<include/device.h> to the ion.h trace file to resolve an undefined
struct warning.
Change-Id: Ic0dedbad134147cc1d02a3d8a8f4fd9a7920b77b
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Fix up a number of headers to work with KERNEL_HEADER_TEST ahead
of turning it on by default.
Change-Id: Ic0dedbaddba2fa62ec60d93bf3e4ef06ededa4cc
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
