Adding a change to check whether the pipe has been
setup or not before going to teardown the pipe.
Change-Id: Ibb408fdf3509facd6c18e3238bdbfd7da0b4a643
Signed-off-by: Jagadeesh Ponduru <quic_jponduru@quicinc.com>
From IPAv4.5 versions IPA_TOS_EQ is considered as not supported
and it made to 0XFF, instead IPA_IS_PURE_ACK is given support.
So, when validating the bitfield equations of them, passing
eq_bitfield[i] will throw shift overflow warning as it takes 0xFF.
So, added change to ensure there are no overflow warnings.
Change-Id: I84384723433706784a4ef21b88f33b036edb0a5d
Signed-off-by: Jagadeesh Ponduru <quic_jponduru@quicinc.com>
Added logic to send QMI_IPA_MOVE_NAT_COMPLETE_IND_V01 only when modem
requests it and hence we initialize nat_move_qmi_disabled with our change
Change-Id: I155ef21726a43a780cd25bd766c3d4cb2a586f84
Fix to NULL terminate the peers list ptr
after freeing it, to get rid of use after
free issue.
Change-Id: Ide9fde9e7648a7af561a5b0ae0fa085810e59ea6
Signed-off-by: Prashanth Reddy Baddam <quic_pbaddam@quicinc.com>
* Changed channel tear down command to match standard naming
* Pretty code formatting
Change-Id: I523b86dd76483f40a686dd3027f2b1febea7c0dc
Signed-off-by: Eliad Ben Yishay <quic_ebenyish@quicinc.com>
* sizeof was used on a pointer rather than actual type
Change-Id: I2a57af6353ce166697af43a35da549532e5ec45c
Signed-off-by: Eliad Ben Yishay <quic_ebenyish@quicinc.com>
After passing ipa_q6_clnt pointer to function, if it freed in
different thread it may lead to NULL pointer access. So adding
check to see passed pointer valid or not.
Change-Id: I22e272ebdecc62756ee140081524ab4efdd3d02a
Signed-off-by: Ashok Vuyyuru <quic_avuyyuru@quicinc.com>
To correct unmap the addresses for first and second
pine, add resource index for second pine instance.
Change-Id: I883a4c45ff516619c834c18e25aec6d7b18b86e2
Signed-off-by: Piyush Dhyani<quic_pdhyani@quicinc.com>
Add Header with len value of 0 is valid.
Removed the part in the test which expect to fail with value of 0.
Signed-off-by: Dor Deri <quic_dderi@quicinc.com>
* 2 packets threshold proved to perform best in terms of throughput
* Changing function name ntn -> ntn3 to stress the exact use case
Change-Id: I251cdff2612c568021e649deb7723708022f2495
Signed-off-by: Eliad Ben Yishay <quic_ebenyish@quicinc.com>
Compile vendor DLKMs only if the system wide flag is true or
if the project specific override flag is set.
Change-Id: I744fce818c59d5072c0e085a41c9225fc317e711
Signed-off-by: Chaitanya Pratapa <quic_cpratapa@quicinc.com>
Modified log type in ipa_fltrt_generate_hw_rule_bdy_5_5 since it is
causing redundant logs
Change-Id: Ia759f17c8f475aa5fb6405880d2a3f224a650c84
Signed-off-by: Abhishek Raghuvanshi <quic_araghuva@quicinc.com>
Adding support to save ipa/gsi ipc logs in minidump.
Change-Id: Ic83b173140aae5c985a497f7333596c540b094ff
Signed-off-by: Ashok Vuyyuru <quic_avuyyuru@quicinc.com>
Created new ipc ipa_clk log for adding more debug prints
for IPA Clocks in which provides the explicit time taken
by IPA ICC Voting for BW Paths and IPA Enable/Disable Clocks.
Change-Id: I53b7b5baba0ff72df286c86a4b45b0f860014474
Signed-off-by: Jagadeesh Ponduru <quic_jponduru@quicinc.com>
When GSI stop channel is failing, & retrying & IPA_HW < 4.0,
the return variable is being updated by the immediate other
function as well, there it results in updating & returning
a wrong value, finally leading to exit without removing endp_delay
on that pipe, results in without stopping the channel thereafter
so corrected it with adding a new variable.
Change-Id: If0ee3986ae40941ce4409d6fb5d74b37db9b7190
Signed-off-by: Jagadeesh Ponduru <quic_jponduru@quicinc.com>
Initial version of minidump support in IPA driver.
Change-Id: I73e98fd647dd487dda90049f5b0c5609f558aa92
Signed-off-by: Ashok Vuyyuru <quic_avuyyuru@quicinc.com>
* In case extension header exists the rule is 2 bytes into the extra words
area
* Subtract extension header size to point to the next whole 8 byte word
right after the header
Change-Id: Ic6e4fc05ca8e03408530cc9815f4ce7455bd7c57
Signed-off-by: Eliad Ben Yishay <quic_ebenyish@quicinc.com>
The change adds Traffic Shaping and Policing (TSP) feature
to the IPA driver:
* HAL addition for TSP tables translation to and from HW format
* Core TSP for table management
* IOCTL handlers for table management commands from userspace
* Debugfs addition that dumps the TSP configuration
The whole feature is added under CONFIG_IPA_TSP kbuild switch.
Change-Id: I11e444aeb93e10d6809873d02a8b3be7cc366755
Signed-off-by: Ilia Lin <quic_ilial@quicinc.com>
Link loading showed a stack based buffer, used for printing
statistics, was too big. The sized has been reduced to be more stack
friendly.
Change-Id: Iab6f6d1d7e77c42855205641c20999c2633f3075
Signed-off-by: Perry Randise <quic_prandise@quicinc.com>
Updated change to avoid race condition and NULL pointer access,
In case of SSR and ioctl call while performing QMI transaction.
Change-Id: I09dbf33d76a3a0d9e4917e62aaf1257a1abe2db9
Signed-off-by: Raghavendar rao l <quic_rlomte@quicinc.com>
In apps resume scenario if start channel not saving IPA registers,
as ipa clock count was not updated. Adding changes to increment clock
count once clock enable success before reg save.
Change-Id: Ie179bfe648373dad0148c65a11f03945292911f5
Signed-off-by: Ashok Vuyyuru <quic_avuyyuru@quicinc.com>
In SSR scenario if QMI response delayed SSR events getting timeout,
To avoid this issue reduce the QMI timeout value from 60sec to 10sec.
Change-Id: Ib0113a91f9bd60b3ea0e374bf249d4859a22a040
Signed-off-by: Ashok Vuyyuru <quic_avuyyuru@quicinc.com>
When commit header table from memory to HW, adding fix to clean
dma allocation if it fails in multiple mapping attempts and also to
clean the allocated mapping of hdr table after commit to HW.
Change-Id: Icbf1b855e34625ffe8f8356a59cddf85a91b0160
Signed-off-by: Jagadeesh Ponduru <quic_jponduru@quicinc.com>
