On 2/22/2024 9:31 PM, Baochen Qiang wrote: > Currently buf_len field of ath11k_mhi_config_qca6390 is assigned > with 0, making MHI use a default size, 64KB, to allocate channel > buffers. This is likely to fail in some scenarios where system > memory is highly fragmented and memory compaction or reclaim is > not allowed. > > There is a fail report which is caused by it: > kworker/u32:45: page allocation failure: order:4, mode:0x40c00(GFP_NOIO|__GFP_COMP), nodemask=(null),cpuset=/,mems_allowed=0 > CPU: 0 PID: 19318 Comm: kworker/u32:45 Not tainted 6.8.0-rc3-1.gae4495f-default #1 openSUSE Tumbleweed (unreleased) 493b6d5b382c603654d7a81fc3c144d59a1dfceb > Workqueue: events_unbound async_run_entry_fn > Call Trace: > <TASK> > dump_stack_lvl+0x47/0x60 > warn_alloc+0x13a/0x1b0 > ? srso_alias_return_thunk+0x5/0xfbef5 > ? __alloc_pages_direct_compact+0xab/0x210 > __alloc_pages_slowpath.constprop.0+0xd3e/0xda0 > __alloc_pages+0x32d/0x350 > ? mhi_prepare_channel+0x127/0x2d0 [mhi 40df44e07c05479f7a6e7b90fba9f0e0031a7814] > __kmalloc_large_node+0x72/0x110 > __kmalloc+0x37c/0x480 > ? mhi_map_single_no_bb+0x77/0xf0 [mhi 40df44e07c05479f7a6e7b90fba9f0e0031a7814] > ? mhi_prepare_channel+0x127/0x2d0 [mhi 40df44e07c05479f7a6e7b90fba9f0e0031a7814] > mhi_prepare_channel+0x127/0x2d0 [mhi 40df44e07c05479f7a6e7b90fba9f0e0031a7814] > __mhi_prepare_for_transfer+0x44/0x80 [mhi 40df44e07c05479f7a6e7b90fba9f0e0031a7814] > ? __pfx_____mhi_prepare_for_transfer+0x10/0x10 [mhi 40df44e07c05479f7a6e7b90fba9f0e0031a7814] > device_for_each_child+0x5c/0xa0 > ? __pfx_pci_pm_resume+0x10/0x10 > ath11k_core_resume+0x65/0x100 [ath11k a5094e22d7223135c40d93c8f5321cf09fd85e4e] > ? srso_alias_return_thunk+0x5/0xfbef5 > ath11k_pci_pm_resume+0x32/0x60 [ath11k_pci 830b7bfc3ea80ebef32e563cafe2cb55e9cc73ec] > ? srso_alias_return_thunk+0x5/0xfbef5 > dpm_run_callback+0x8c/0x1e0 > device_resume+0x104/0x340 > ? __pfx_dpm_watchdog_handler+0x10/0x10 > async_resume+0x1d/0x30 > async_run_entry_fn+0x32/0x120 > process_one_work+0x168/0x330 > worker_thread+0x2f5/0x410 > ? __pfx_worker_thread+0x10/0x10 > kthread+0xe8/0x120 > ? __pfx_kthread+0x10/0x10 > ret_from_fork+0x34/0x50 > ? __pfx_kthread+0x10/0x10 > ret_from_fork_asm+0x1b/0x30 > </TASK> > > Actually those buffers are used only by QMI target -> host communication. > And for WCN6855 and QCA6390, the largest packet size for that is less > than 6KB. So change buf_len field to 8KB, which results in order 1 > allocation if page size is 4KB. In this way, we can at least save some > memory, and as well as decrease the possibility of allocation failure > in those scenarios. > > Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.30 > > Reported-by: Vlastimil Babka <vbabka@xxxxxxx> > Closes: https://lore.kernel.org/ath11k/96481a45-3547-4d23-ad34-3a8f1d90c1cd@xxxxxxx/ > Signed-off-by: Baochen Qiang <quic_bqiang@xxxxxxxxxxx> Acked-by: Jeff Johnson <quic_jjohnson@xxxxxxxxxxx>
