In the 'commit 
ebee76f7fa46 ("ath10k: allow setting coverage class")',
it inherits the design and the address offset from ath9k, but the address
is not applicable to QCA6174, which leads to a random crash while doing the
resume() operation, since the set_coverage_class.ops will be called from
ieee80211_reconfig() when resume() (if the wow is not configured).
Fix the incorrect address offset here to avoid the random crash.
Verified on QCA6174/hw3.0 with firmware WLAN.RM.4.4-00022-QCARMSWPZ-2.
kvalo: this also seems to fix a regression with firmware restart.
Fixes: ebee76f7fa46 ("ath10k: allow setting coverage class")
Cc: <stable@vger.kernel.org> # v4.10
Signed-off-by: Ryan Hsu <ryanhsu@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
        .rtc_soc_base_address                   = 0x00000800,
        .rtc_wmac_base_address                  = 0x00001000,
        .soc_core_base_address                  = 0x0003a000,
-       .wlan_mac_base_address                  = 0x00020000,
+       .wlan_mac_base_address                  = 0x00010000,
        .ce_wrapper_base_address                = 0x00034000,
        .ce0_base_address                       = 0x00034400,
        .ce1_base_address                       = 0x00034800,