Bluetooth: btrtl: Use kvmalloc for FW allocations
PATCH 1/1 Bluetooth: btrtl: Use kvmalloc for FW allocations
Currently, kmemdup is applied to the firmware data, and it invokes kmalloc under the hood. The firmware size and patch_length are big (more than PAGE_SIZE), and on some low-end systems (like ASUS E202SA) kmalloc may fail to allocate a contiguous chunk under high memory usage and fragmentation:
Bluetooth: hci0: RTL: examining hci_ver=06 hci_rev=000a lmp_ver=06 lmp_subver=8821
Bluetooth: hci0: RTL: rom_version status=0 version=1
Bluetooth: hci0: RTL: loading rtl_bt/rtl8821a_fw.bin
kworker/u9:2: page allocation failure: order:4, mode:0x40cc0(GFP_KERNEL|__GFP_COMP), nodemask=(null),cpuset=/,mems_allowed=0 <stack trace follows>
As firmware load happens on each resume, Bluetooth will stop working after several iterations, when the kernel fails to allocate an order-4 page.
This patch replaces kmemdup with kvmalloc+memcpy. It's not required to have a contiguous chunk here, because it's not mapped to the device directly.
--- drivers/bluetooth/btrtl.c | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-)
drivers/bluetooth/btrtl.c +11 -9
@@ -370,11 +370,11 @@ static int rtlbt_parse_firmware(struct hci_dev *hdev,* the end. */ len = patch_length; - buf =kmemdup (btrtl_dev->fw_data + patch_offset, patch_length, -GFP_KERNEL ); + buf =kvmalloc (patch_length,GFP_KERNEL );if (!buf)return -ENOMEM ; +memcpy (buf, btrtl_dev->fw_data + patch_offset, patch_length -4 );memcpy (buf + patch_length -4 , &epatch_info->fw_version,4 ); *_buf = buf;
@@ -460,8 +460,10 @@ static int rtl_load_file(struct hci_dev *hdev, const char *name, u8 **buff)if (ret <0 )return ret; ret = fw->size; - *buff =kmemdup (fw->data, ret,GFP_KERNEL ); -if (!*buff) + *buff =kvmalloc (fw->size,GFP_KERNEL ); +if (*buff) +memcpy (*buff, fw->data, ret); +else ret = -ENOMEM ;release_firmware (fw);
@@ -499,14 +501,14 @@ static int btrtl_setup_rtl8723b(struct hci_dev *hdev, goto out;if (btrtl_dev->cfg_len >0 ) { - tbuff =kzalloc (ret + btrtl_dev->cfg_len,GFP_KERNEL ); + tbuff =kvzalloc (ret + btrtl_dev->cfg_len,GFP_KERNEL );if (!tbuff) { ret = -ENOMEM ; goto out; }memcpy (tbuff, fw_data, ret); -kfree (fw_data); +kvfree (fw_data);memcpy (tbuff + ret, btrtl_dev->cfg_data, btrtl_dev->cfg_len); ret += btrtl_dev->cfg_len;
@@ -519,14 +521,14 @@ static int btrtl_setup_rtl8723b(struct hci_dev *hdev, ret =rtl_download_firmware (hdev, fw_data, ret); out: -kfree (fw_data); +kvfree (fw_data);return ret; }void btrtl_free (struct btrtl_device_info *btrtl_dev) { -kfree (btrtl_dev->fw_data); -kfree (btrtl_dev->cfg_data); +kvfree (btrtl_dev->fw_data); +kvfree (btrtl_dev->cfg_data);kfree (btrtl_dev); }EXPORT_SYMBOL_GPL (btrtl_free);
Signature · 1 line
2.25.0
Marcel Holtmann
maintainerHi Maxim,
Quoted 16 lines
Currently, kmemdup is applied to the firmware data, and it invokes kmalloc under the hood. The firmware size and
patch_lengthare big (more thanPAGE_SIZE), and on some low-end systems (like ASUS E202SA) kmalloc may fail to allocate a contiguous chunk under high memory usage and fragmentation:Bluetooth: hci0: RTL: examining hci_ver=06 hci_rev=000a lmp_ver=06 lmp_subver=8821
Bluetooth: hci0: RTL:rom_versionstatus=0 version=1
Bluetooth: hci0: RTL: loading rtl_bt/rtl8821a_fw.bin
kworker/u9:2: page allocation failure: order:4, mode:0x40cc0(GFP_KERNEL|__GFP_COMP), nodemask=(null),cpuset=/,mems_allowed=0 <stack trace follows>As firmware load happens on each resume, Bluetooth will stop working after several iterations, when the kernel fails to allocate an order-4 page.
Quoted 8 lines
This patch replaces kmemdup with kvmalloc+memcpy. It's not required to have a contiguous chunk here, because it's not mapped to the device directly.
Signed-off-by Maxim Mikityanskiy <[email protected]>---
drivers/bluetooth/btrtl.c| 20 +++++++++++---------1 file changed, 11 insertions(+), 9 deletions(-)
patch has been applied to bluetooth-next tree.
Regards
Marcel