On Wed, Feb 03, 2021 at 10:14:01AM +0800, Mingchuang Qiao wrote: > Hi, > > On Mon, 2021-02-01 at 13:32 +0200, Mika Westerberg wrote: > > Hi, > > > > On Fri, Jan 29, 2021 at 03:11:37PM +0800, mingchuang.qiao@xxxxxxxxxxxx wrote: > > > From: Mingchuang Qiao <mingchuang.qiao@xxxxxxxxxxxx> > > > > > > In bus scan flow, the "LTR Mechanism Enable" bit of DEVCTL2 register is > > > configured in pci_configure_ltr(). If device and bridge both support LTR > > > mechanism, the "LTR Mechanism Enable" bit of device and bridge will be > > > enabled in DEVCTL2 register. And pci_dev->ltr_path will be set as 1. > > > > > > If PCIe link goes down when device resets, the "LTR Mechanism Enable" bit > > > of bridge will change to 0 according to PCIe r5.0, sec 7.5.3.16. However, > > > the pci_dev->ltr_path value of bridge is still 1. > > > > > > For following conditions, check and re-configure "LTR Mechanism Enable" bit > > > of bridge to make "LTR Mechanism Enable" bit mtach ltr_path value. > > > > Typo mtach -> match. > > > > > -before configuring device's LTR for hot-remove/hot-add > > > -before restoring device's DEVCTL2 register when restore device state > > > > > > Signed-off-by: Mingchuang Qiao <mingchuang.qiao@xxxxxxxxxxxx> > > > --- > > > changes of v2 > > > -modify patch description > > > -reconfigure bridge's LTR before restoring device DEVCTL2 register > > > changes of v3 > > > -call pci_reconfigure_bridge_ltr() in probe.c > > > > Hmm, which part of this patch takes care of the reset path? It is not > > entirely clear to me at least. > > > > When device resets and link goes down, there seems to have two methods > to recover for software. > -One is that trigger device removal and rescan. > -The other is that restore device with pci_restore_state() after link > comes back up. > For above both scenarios, we need check and reconfigure "LTR Mechanism > Enable" bit of bridge. It's also this patch intends to do. > -For the rescan scenario, it's done in pci_configure_ltr(). > -For the restore scenario, it's done in pci_restore_pcie_state(). Okay, thanks for the clarification!
