| Summary: | Large delay in activating network for RTL8111 | ||
|---|---|---|---|
| Product: | Mageia | Reporter: | Frank Griffin <ftg> |
| Component: | Installer | Assignee: | Mageia Bug Squad <bugsquad> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Normal | CC: | cjw, thierry.vignaud, tmb |
| Version: | Cauldron | Keywords: | NEEDINFO |
| Target Milestone: | --- | ||
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Source RPM: | drakx-installer-binaries | CVE: | |
| Status comment: | |||
| Attachments: | reprot.bug.gz using the RTL8111 NIC | ||
|
Description
Frank Griffin
2012-12-20 18:04:55 CET
Created attachment 3263 [details]
reprot.bug.gz using the RTL8111 NIC
Maybe firmware loading. The following message appears one minute after loading previous modules "unable to load firmware patch rtl_nic/rtl8168e-3.fw (-2)" Did you use boot-nonfree.iso or boot.iso? CC:
(none) =>
thierry.vignaud, tmb I used boot.iso. But I don't understand two things: 1) Why it eventually works just fine without firmware anyway. 2) Why it takes a minute to decide it can't find firmware. 1) maybe only some features needs a firmware 2) a timeout (In reply to comment #4) > 1) maybe only some features needs a firmware > 2) a timeout Yeah, but in either case, either the package is there or it isn't. If it isn't, it isn't going to magically appear because you wait for a minute. Indeed. Anyway does boot-nonfree.iso works better? I'll try that today. With boot-nonfree, both NICs bring up the network immediately. Trying again with boot.iso, the onboard NIC still gets the delay, but the card is fine. The interesting thing is that these are the same NICs from bug#8451, which are supposed to be identical. You could try removing /sbin/hotplug and /hotplug from all.rdz in order to see if it's it that causes the timeout issue
MGA_PATH=/mageia/unstable/x86_64/isolinux/alt0
mkdir T
cd T
xzcat $MGA_PATH/all.rdz|cpio -id
rm -f hotplug sbin/hotplug
cp $MGA_PATH/all.rdz{,.sav}
find| cpio -o -c --quiet| xz --check=crc32 --lzma2=dict=512KiB>$MGA_PATH/all.rdz
Anyway, I think I'll just remove our hotplug since kernel now autoloads firmware. Thomas: no complaint against this? There are likely different versions of these chips, lspci says 8168B while the firmware files are for 8168d,e,f,g - so maybe the driver does not load a firmware file for some versions. For example I have "b" and "e" chips (plus a 8110s) while you have "b" and "evl" versions (according to the driver's dmesg output). CC:
(none) =>
cjw >There are likely different versions of these chips
Makes sense. But it still doesn't explain why it takes the code 60 seconds or more to figure out the firmware package isn't there and move on.
Does it works better with mga4 with regular boot.iso? We now rely on kernel for loading firmware. Keywords:
(none) =>
NEEDINFO @ftg@roadrunner.com: ping? This was reported against an older installer, no longer current. Please revert to #c14 from tv, but bearing in mind to base a reply on M5beta1 If okay please close. If not please advise, remove the 'NEEDINFO' and set whiteboard as '5beta1' (or higher) . Thanks! I'm closing this. Looking at harddrake2 on the affected box, it shows as an NC100 Network Everywhere Fast Ethernet 10/100; it's the PCI-E card that is the RTL8111 Gigabit. Since I've upgraded the rest of my network to gigabit, I've disabled the NC100 in the BIOS, so this is really no longer an issue. Status:
NEW =>
RESOLVED |