Bug 194801
| Summary: | e1000e: __pci_enable_msi_range fails before/after resume | ||
|---|---|---|---|
| Product: | Drivers | Reporter: | Rouven Czerwinski (rouven+kbugzilla) |
| Component: | Network | Assignee: | drivers_network (drivers_network) |
| Status: | RESOLVED CODE_FIX | ||
| Severity: | normal | CC: | regressions, shawn.starr, yu.c.chen |
| Priority: | P1 | ||
| Hardware: | x86-64 | ||
| OS: | Linux | ||
| Kernel Version: | 4.11-rc1 | Subsystem: | |
| Regression: | No | Bisected commit-id: | |
| Attachments: | dmesg 4.11-rc1 | ||
Commit 7e54d9d063fa239c95c21548c5267f0ef419ff56 moved the e1000e_reset_interrupt_capability(adapter) call into the netif_running conditional, meaning the interrupts will only be reset if the network interface is up, which is not necessarily the case on my Laptop and pm_freeze is called to suspend it. Reverting the commit fixes my issue. *** Bug 194807 has been marked as a duplicate of this bug. *** @Chen Yu, others: What's the status of this? are the e1000e developers aware of this? Did anyone brought it forward on their list and/or the network developers? JFYI: I added this report to the list of regressions for Linux 4.11. I'll watch this place for further updates on this issue to document progress in my weekly reports, hence to allow me to keep track please send a quick note here in case this issue gets discussed somewhere else. There is an ongoing discussion on LKML starting with this message: https://lkml.org/lkml/2017/3/12/6 The tester of the above mentioned commit posted to this thread as well. This is still not fixed in -rc5, any chance this will be reverted before 4.11 is released? The revert is already sitting in davems tree, see https://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git/commit/drivers/net/ethernet/intel/e1000e/netdev.c?id=9f47a48e6eb97d793db85373e3ef4c55d876d334, we'll just have to wait for the next net pull, which will surely happen before the 4.11 release. The revert hit Linus tree with the pull from daves net tree, closing this issue. |
Created attachment 255101 [details] dmesg 4.11-rc1 On 4.11-rc1 the whole ethernet subsystem seems to hang after a suspend/resume cycle, see attached dmesg.