Includes changes from: https://github.com/pershoot/gtab-2632/commits/gingerbread/drivers/net/wireless/bcm4329 Plus modifications for Huawei U8160 support
From: https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm.git;a=commit;h=74394aee55bb2d92d05ad5cd373d8e553e101bee [PATCH] msm: io: Shift up device iomaps to increase virtual address range Shift up all the virtual addresses for the device iomaps by 256M so that this range is free for use by the system. From: https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm.git;a=commit;h=7bf9edd9cf4bec4029f5438dbb13cfc4f3674753 [PATCH] msm: Shift-up device iomaps and vmalloc area Shifting these up increases the virtual address range available for the kernel, which can be either used for vmalloc space or the 1-1 mapping depending upon the target.
EOS buffer is not getting processed in following scenario is encountered: -Driver Recieves FLUSH command from apps -Application sends EOS buffer to the driver after flush command has completed -NEEDS_DATA messages from DSP is Recieved by driver In above scenario driver blocks for forever in write wait queue and EOS buffer never gets processed. Fix this by sending wakeup when NEEDS_DATA message is received. CRs-Fixed: 292954, 293145 Signed-off-by: Manish Dewangan <firstname.lastname@example.org> (cherry picked from commit c729e63e1adc0b1e3b542477731e3485574150cd) Change-Id: Iba4f7850e546d3816c31389a33984af28cca6503 Signed-off-by: Sunil Joseph <email@example.com>
Audio drivers does not release spinlock when kmalloc() fails in post_event function. This can result in deadlock or system hang in low memory conditions. Fix this bug by releasing the spinlock when kmalloc fails. Change-Id: Id6ffd5c34ac36f68cb7b490a43bdf09bb80c5b9e CRs-Fixed: 281588 Signed-off-by: Sriranjan Srikantam <firstname.lastname@example.org>
This change is to support Real Video decoder. 0x9201 is the frame header ID for Real decoder. Slice data has YUV buffer addresses where DSP has to write decoded data, but these addresses are virtual addresses. This code is added to convert virtual addresses to physical addresses before passing them to DSP. Change-Id: Ib129047291558d55708d31994f723ac332ce5092 Signed-off-by: Mahesh Lanka <email@example.com>
When DSP is not ready to accept a command, ADSP driver retries for 100 times with 50 usecs delay each time which amounts to 5 msec in total. But when DSP is busy processing commands it may take more than 5 msec to respond to requests from application processor. Hence updating the number of trials to 300 with 50 usecs delay each time thus increasing the maximum delay to 15 msecs. Change-Id: I641d1f09c232621823217a8a9f2754676467c319 Signed-off-by: Suresh Vankadara <firstname.lastname@example.org>
This change is to support H264 B frame clips. 0x8201 is the frame header ID for H264 VLD in ARM path. Slice data has YUV buffer addresses where DSP has to write decoded data, but these addresses are virtual addresses. This code is added to convert virtual addresses to physical addresses before passing them to DSP Change-Id: I287a6adbecf27e02c0c77a8078208ddc474f6b48 Signed-off-by: Mahesh Lanka <email@example.com>
DSP restricts input bitstream buffer size to be minimum of 8KB for wmapro decoder. Hence increase the input buffer size value. Change-Id: I4f04ad9b2d7ecc41fdd67ff7bfce1252d4b00af3 CRs-Fixed: 260639 Signed-off-by: Sriranjan Srikantam <firstname.lastname@example.org>
This reverts commit b38db38.
This reverts commit d944b65.
The lowmemorykiller registers an atomic notifier for notfication of when the task is freed. From this atomic notifier callback, it removes the atomic notifier via task_free_unregister(). This is incorrect because atomic_notifier_chain_unregister() calls syncronize_rcu(), which can sleep, which shouldn't be done from an atomic notifier. Fix this by registering the notifier during init, and only unregister it if the lowmemorykiller is unloaded. Change-Id: Ie9cf2f23d31c78e1fe1b9dd2e11af5267832ee94 Signed-off-by: Rabin Vincent <email@example.com> Signed-off-by: Christian Bejram <firstname.lastname@example.org>
…r a process to die If a process forked and the child process was killed by the lowmemorykiller, the lowmemory killer would be disabled until the parent process reaped the child or it died itself. Change-Id: I49061da80aeae94aa8a244c1cc48b80bef6dc446 Signed-off-by: Arve HjÃ¸nnevÃ¥g <email@example.com>