Bug 83631
Summary: | RTL8187SE Performance issues | ||
---|---|---|---|
Product: | Drivers | Reporter: | ZED (ghostz191989) |
Component: | network-wireless | Assignee: | drivers_network-wireless (drivers_network-wireless) |
Status: | NEW --- | ||
Severity: | high | CC: | andrea.merello |
Priority: | P1 | ||
Hardware: | x86-64 | ||
OS: | Linux | ||
Kernel Version: | 3.16 | Subsystem: | |
Regression: | No | Bisected commit-id: | |
Attachments: |
rtl818x cumulative for 3.16
Fix response rate patch. |
Description
ZED
2014-09-01 01:15:18 UTC
The signal strength indication in rtl818x_pci driver is not enabled yet for rtl8187se. It always reports a dummy number regardless the real signal strength. I have a patch that adds support for it. This has been not merged in the driver yet because I need someone that tests it (but I'm quite confident it should -at least rougly- work). Cold you please do that for me ? http://marc.info/?l=linux-wireless&m=140155388332534&w=2 Please don't expect the same signal strength reported by the older driver: They are probably not is the same scale. Just to give you an idea, I expect the reported signal to be about -80dBm to -90dBm when the signal is very poor, rising to something like -40dBm when the signal is very strong. Despite the stuck signal indicator may suggest "the thing is bad" at the very first look, do you confirm that the performance problem has really exists on both 3.15 and 3.16 ? Can you please check, and report, the link rate with the old (3.14) driver and with the newer one (I mean the reported link speed, not the measured throughput) ? Can you please try to force the link rate on the newer driver to the same link rate you get using the older driver, and check if the actual performances of the new driver increase this way? I used your patch with patch command for 3.15 and 3.16 and the result is this patching file drivers/net/wireless/rtl818x/rtl8180/dev.c Hunk #3 succeeded at 232 with fuzz 2. Hunk #4 FAILED at 241. Hunk #5 FAILED at 279. Hunk #6 FAILED at 289. Hunk #7 FAILED at 1823. 4 out of 7 hunks FAILED -- saving rejects to file drivers/net/wireless/rtl818x/rtl8180/dev.c.rej patch unexpectedly ends in middle of line The Link speed is the same with the old and the new driver. Created attachment 151891 [details]
rtl818x cumulative for 3.16
Sorry, the former patch was intended for wireless-testing tree. I just added a cumulative patch, that you may apply to your 3.16 kernel. This contains latest fixes and should make you able to apply the RSSI patch on top of it. Unfortunately I got an error also with this patch. https://bpaste.net/show/d1577570eb61 Forgot to mention that I was trying to apply it to 3.16 I was try also with 3.17 your RSSI patch seems to work with 77% signal strength while the old driver from 3.14 shows 85% in the same environment. The performance is still poor. About the signal strength, difference between old and new driver could be ok: this is probably just due to different measure unit in reported signal strength. If you run iw wlan0 link (assuming your card is wlan0) you should get measurement in absolute dBm unit. About performance I have two new patches. One has been just merged in wireless testing tree (possibly it may appply on top of your 3.17 + rssi patch kernel) and the other is on my PC waiting for me having some time to check it once more time.. Possibly next days I will see if the former patch applies on the said kernel, otherwise I will post some rebased version here Created attachment 153251 [details]
Fix response rate patch.
This patch should apply on top of: 3.17+rssi_patch
I have apply the latest patch you gave me. The results are much better almost the same with the old driver. old driver(3.14.22 Kernel): 9.26 Mbps Throughput new driver + your patches(3.17.1 Kernel): 9.03 Mbps Throughput The packet loss is the same 1% both. I will try to test with another AP because the packet loss is maybe a problem from the current AP. Thank you very much and keep up the good work. Hello again. When this driver is gonna be in the kernel officially? |