Re: [PATCH v3 2/3] uio_pci_generic: add MSI/MSI-X support

From: Vlad Zolotarov
Date: Mon Oct 05 2015 - 06:06:17 EST




On 10/05/15 12:08, Vlad Zolotarov wrote:


On 10/05/15 11:41, Stephen Hemminger wrote:
On Sun, 4 Oct 2015 23:43:17 +0300
Vlad Zolotarov <vladz@xxxxxxxxxxxxxxxxxxxx> wrote:

+static int setup_maps(struct pci_dev *pdev, struct uio_info *info)
+{
+ int i, m = 0, p = 0, err;
+ static const char * const bar_names[] = {
+ "BAR0", "BAR1", "BAR2", "BAR3", "BAR4", "BAR5",
+ };
+
+ for (i = 0; i < ARRAY_SIZE(bar_names); i++) {
+ unsigned long start = pci_resource_start(pdev, i);
+ unsigned long flags = pci_resource_flags(pdev, i);
+ unsigned long len = pci_resource_len(pdev, i);
+
+ if (start == 0 || len == 0)
+ continue;
+
+ if (flags & IORESOURCE_MEM) {
+ void __iomem *addr;
+
+ if (m >= MAX_UIO_MAPS)
+ continue;
+
+ addr = ioremap(start, len);
+ if (addr == NULL) {
+ err = -EINVAL;
+ goto fail;
+ }
+
+ info->mem[m].name = bar_names[i];
+ info->mem[m].addr = start;
+ info->mem[m].internal_addr = addr;
+ info->mem[m].size = len;
+ info->mem[m].memtype = UIO_MEM_PHYS;
+ ++m;
+ } else if (flags & IORESOURCE_IO) {
+ if (p >= MAX_UIO_PORT_REGIONS)
+ continue;
+
+ info->port[p].name = bar_names[i];
+ info->port[p].start = start;
+ info->port[p].size = len;
+ info->port[p].porttype = UIO_PORT_X86;
+ ++p;
+ }
+ }
+
+ return 0;
+fail:
+ for (i = 0; i < m; i++) {
+ iounmap(info->mem[i].internal_addr);
+ info->mem[i].internal_addr = NULL;
+ }
+
+ return err;
+
I wonder do we really have to setup all the BAR's in uio_pci_generic?
The DPDK code works with uio_pci_generic already, and it didn't setup the BAR's.

DPDK never used uio_pci_generic with MSI-X support so far and all MSI-X capable UIO DPDK drivers like igb_uio and your newly proposed uio_msi do map them all.
U also mentioned in the other thread that virtio requires portio bars too.
The thing is that generally bars are needed for programming the device therefore it's logical to have them exposed. In general different devices have different registers layout therefore a general driver may not know which bar exactly is going to be required for a specific device and for a specific usage. That's why I think that "map them all" approach is rather generic and appropriate. However if there are other motives not to do so that I'm missing, pls., let me know.

Having said all that however I'd agree if someone would say that mappings setting would rather come as a separate patch in this series... ;)
it will in v4...



thanks,
vlad

One possible issue is that without that maybe kernel would not know about the
region used for MSI-X vectors table.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/