On 16-06-22, 18:20, Serge Semin wrote: > On Thu, Jun 16, 2022 at 07:30:23AM -0700, Vinod Koul wrote: > > On 16-06-22, 17:03, Serge Semin wrote: > > > On Thu, Jun 16, 2022 at 05:02:00PM +0300, Serge Semin wrote: > > > > On Thu, Jun 16, 2022 at 04:54:13PM +0300, Serge Semin wrote: > > > > > On Thu, Jun 16, 2022 at 06:51:55AM -0700, Vinod Koul wrote: > > > > > > On 24-05-22, 10:21, Frank Li wrote: > > > > > > > Default Designware EDMA just probe remotely at host side. > > > > > > > This patch allow EDMA driver can probe at EP side. > > > > > > > > > > > > > > 1. Clean up patch > > > > > > > dmaengine: dw-edma: Detach the private data and chip info structures > > > > > > > dmaengine: dw-edma: Remove unused field irq in struct dw_edma_chip > > > > > > > dmaengine: dw-edma: Change rg_region to reg_base in struct > > > > > > > dmaengine: dw-edma: rename wr(rd)_ch_cnt to ll_wr(rd)_cnt in struct > > > > > > > > > > > > > > 2. Enhance EDMA driver to allow prode eDMA at EP side > > > > > > > dmaengine: dw-edma: Add support for chip specific flags > > > > > > > dmaengine: dw-edma: Add DW_EDMA_CHIP_32BIT_DBI for chip specific > > > > > > > flags (this patch removed at v11 because dma tree already have fixed > > > > > > > patch) > > > > > > > > > > > > > > 3. Bugs fix at EDMA driver when probe eDMA at EP side > > > > > > > dmaengine: dw-edma: Fix programming the source & dest addresses for > > > > > > > ep > > > > > > > dmaengine: dw-edma: Don't rely on the deprecated "direction" member > > > > > > > > > > > > > > 4. change pci-epf-test to use EDMA driver to transfer data. > > > > > > > PCI: endpoint: Add embedded DMA controller test > > > > > > > > > > > > > > 5. Using imx8dxl to do test, but some EP functions still have not > > > > > > > upstream yet. So below patch show how probe eDMA driver at EP > > > > > > > controller driver. > > > > > > > https://lore.kernel.org/linux-pci/20220309120149.GB134091@thinkpad/T/#m979eb506c73ab3cfca2e7a43635ecdaec18d8097 > > > > > > > > > > > > > > > > > Applied to dmaengine-next, thanks > > > > > > > > > > > > > > Vinod, this was supposed to be merged in through PCIe repo.( I asked > > > > > many times of that. Bjorn also agreed to merge it in. Could drop it > > > > > from yout repo? > > > > > > > > > > > I asked it several time including in the framework of this thread: > > > > https://lore.kernel.org/dmaengine/20220525092306.wuansog6fe2ika3b@mobilestation/ > > > > There are dependencies of my patchsets from this one. Please consider > > > > dropping it from your dmaengine-next repo while it's still possible > > > > since taking DW eDMA and PCie patches through the PCie repo would be > > > > more natural. > > > > > > The only thing we were waiting for was you ack tag... > > > > > Is there any dependency, since this has many dmaengine patches and one > > pcie patch which was acked by Kishon, so can go thru dmaengine tree... > > Right. This one is fully suitable for the DMA-engine tree, but there > are dependencies of my work from this series. My patchsets in its turn > concern both PCIe and DMA-engine parts. > > Here is the deal. My work consists of four patchsets: > [PATCH v4 00/18] PCI: dwc: Various fixes and cleanups > https://lore.kernel.org/linux-pci/20220610082535.12802-1-Sergey.Semin@xxxxxxxxxxxxxxxxxxxx/ > [PATCH v3 00/15] PCI: dwc: Add hw version and dma-ranges support > https://lore.kernel.org/linux-pci/20220610084444.14549-1-Sergey.Semin@xxxxxxxxxxxxxxxxxxxx/ > [PATCH v3 00/17] PCI: dwc: Add generic resources and Baikal-T1 support > https://lore.kernel.org/linux-pci/20220610085706.15741-1-Sergey.Semin@xxxxxxxxxxxxxxxxxxxx/ > [PATCH v3 00/24] dmaengine: dw-edma: Add RP/EP local DMA controllers support > https://lore.kernel.org/dmaengine/20220610091459.17612-1-Sergey.Semin@xxxxxxxxxxxxxxxxxxxx/ > > The last patchset depends on the Frank series and my DW PCIe > patchsets listed above. To cut it shortly here is the > dependencies graph: > > PCIe RP/EP | DMA engine > ^ | ^ > | | | > 1. Semin: [PATCH v4 00/18] "PCI: dwc: | | > Various fixes and cleanups" | | > ^ | | > | | | > 2. Semin: [PATCH v3 00/15] "PCI: dwc: | | > Add hw version and dma-ranges | | > support" | | > ^ | | > | | | > 3. Semin: [PATCH v3 00/17] "PCI: dwc: | 4. Li: [PATCH v12 0/8] "Enable > Add generic resources and Baikal-T1| designware PCI EP EDMA locally" > support" | ^ > ^ | > | | > +-----------------+-------------------+ > | > 5. Semin: [PATCH v3 00/24] "dmaengine: dw-edma: > Add RP/EP local DMA controllers support" > > As you can see my series (5) depends on the Frank' series (4). My series > in its turn depends on the PCI patchsets (3), (2) and (1). So if you get > to merge the Frank' patchset into your tree we'll need to merge the > rest of the patchsets through your repo too. Seeing the DW eDMA device > is implemented as either a separate PCIe device or embedded into the > DW PCIe RP/EP controller and there are many PCI-related patches in my > work, it would be more suitable to merge all the patches through the > PCIe tree. Bjorn already agreed to do that here: > https://lore.kernel.org/linux-pci/20220524155201.GA247821@bhelgaas/ > We only waited for your ackes. BTW should you have some free time > please review my series (5) too since it mainly concerns the DW eDMA > driver (though Mani, in Cc, has already done it and even performed > the testing). > > > > > I can drop if that suits all... > > Yes please. Please drop the Frank' series from your repo, give your > explicit ack if you are ok with it and let's merge it and the rest of > the work through the Bjorn' repo. Patchset (1) has been fully reviewed. > Series (2) and (3) need one more iteration to be finished. So I very > much hope they will be done in the next few weeks, before the next > merge window. Ok dropped now -- ~Vinod