From: Filip Jensen <dev-Felipe.Jensen@duagon.com>
To: morbidrsa@gmail.com
Cc: linux-kernel@vger.kernel.org,
Filip Jensen <dev-Felipe.Jensen@duagon.com>,
Jose Javier Rodriguez Barbarin
<dev-josejavier.rodriguez@duagon.com>
Subject: [PATCH 2/3] mcb: Fix detection of iomapped pci devices
Date: Thu, 26 Feb 2026 16:13:38 +0100 [thread overview]
Message-ID: <20260226151339.48800-3-dev-Felipe.Jensen@duagon.com> (raw)
In-Reply-To: <20260226151339.48800-1-dev-Felipe.Jensen@duagon.com>
The mcb driver in a PCI bus is not detecting if an attached device has
an iomapped or a memory mapped BAR, as all the addresses returned by
pci_resource_start end in 0.
Instead of the low level approach, which is still valid for lpc buses,
here it will be used the standard PCI helper pci_resource_flags for
checking if the PCI BAR is IO mapped (as it is done in mcb_pci_probe)
Reviewed-by: Jose Javier Rodriguez Barbarin <dev-josejavier.rodriguez@duagon.com>
Signed-off-by: Filip Jensen <dev-Felipe.Jensen@duagon.com>
---
drivers/mcb/mcb-parse.c | 25 +++++++++++++++++--------
1 file changed, 17 insertions(+), 8 deletions(-)
diff --git a/drivers/mcb/mcb-parse.c b/drivers/mcb/mcb-parse.c
index 806be67b960d..251be2428873 100644
--- a/drivers/mcb/mcb-parse.c
+++ b/drivers/mcb/mcb-parse.c
@@ -29,9 +29,9 @@ static int chameleon_parse_bdd(struct mcb_bus *bus,
return 0;
}
-static int chameleon_parse_gdd(struct mcb_bus *bus,
- struct chameleon_bar *cb,
- void __iomem *base, int bar_count)
+static int chameleon_parse_gdd(struct mcb_bus *bus, struct chameleon_bar *cb,
+ void __iomem *base, int bar_count,
+ bool mcb_is_pci_device)
{
struct chameleon_gdd __iomem *gdd =
(struct chameleon_gdd __iomem *) base;
@@ -76,8 +76,12 @@ static int chameleon_parse_gdd(struct mcb_bus *bus,
ret = 0;
goto err;
}
-
- if (dev_mapbase & 0x01) {
+ if (mcb_is_pci_device) {
+ if (pci_resource_flags(to_pci_dev(bus->carrier), mdev->bar) &
+ IORESOURCE_IO)
+ goto io_unsupported;
+ } else if (dev_mapbase & 0x01) {
+io_unsupported:
pr_info("IO mapped Device (16z%03d) not yet supported\n",
mdev->id);
ret = 0;
@@ -125,7 +129,7 @@ static void chameleon_parse_bar(void __iomem *base,
}
static int chameleon_get_bar(void __iomem **base, struct chameleon_bar **cb,
- struct device *dev)
+ struct device *dev, bool *mcb_is_pci_device)
{
struct chameleon_bar *c;
int bar_count;
@@ -154,6 +158,7 @@ static int chameleon_get_bar(void __iomem **base, struct chameleon_bar **cb,
chameleon_parse_bar(*base, c, bar_count);
*base += BAR_DESC_SIZE(bar_count);
+ *mcb_is_pci_device = false;
} else {
pdev = to_pci_dev(dev);
bar_count = PCI_STD_NUM_BARS;
@@ -164,6 +169,7 @@ static int chameleon_get_bar(void __iomem **base, struct chameleon_bar **cb,
c[i].addr = pci_resource_start(pdev, i);
c[i].size = pci_resource_len(pdev, i);
}
+ *mcb_is_pci_device = true;
}
*cb = c;
@@ -182,6 +188,7 @@ int chameleon_parse_cells(struct mcb_bus *bus, void __iomem *base)
int ret;
u32 hsize;
u32 table_size;
+ bool mcb_is_pci_device;
hsize = sizeof(struct chameleon_fpga_header);
@@ -207,7 +214,8 @@ int chameleon_parse_cells(struct mcb_bus *bus, void __iomem *base)
memcpy(bus->name, header->filename, CHAMELEON_FILENAME_LEN);
bus->name[CHAMELEON_FILENAME_LEN] = '\0';
- bar_count = chameleon_get_bar(&p, &cb, bus->carrier);
+ bar_count =
+ chameleon_get_bar(&p, &cb, bus->carrier, &mcb_is_pci_device);
if (bar_count < 0) {
ret = bar_count;
goto free_header;
@@ -216,7 +224,8 @@ int chameleon_parse_cells(struct mcb_bus *bus, void __iomem *base)
for_each_chameleon_cell(dtype, p) {
switch (dtype) {
case CHAMELEON_DTYPE_GENERAL:
- ret = chameleon_parse_gdd(bus, cb, p, bar_count);
+ ret = chameleon_parse_gdd(bus, cb, p, bar_count,
+ mcb_is_pci_device);
if (ret < 0)
goto free_bar;
p += sizeof(struct chameleon_gdd);
--
2.34.1
next prev parent reply other threads:[~2026-02-26 15:14 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-26 15:13 [PATCH 0/3] mcb: enable register PCI devices using any BAR Filip Jensen
2026-02-26 15:13 ` [PATCH 1/3] mcb: Use more than the first bar in pci devices Filip Jensen
2026-02-26 15:13 ` Filip Jensen [this message]
2026-02-27 13:57 ` [PATCH 2/3] mcb: Fix detection of iomapped " Johannes Thumshirn
2026-02-26 15:13 ` [PATCH 3/3] mcb: refactor the bus operations out of the common code Filip Jensen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260226151339.48800-3-dev-Felipe.Jensen@duagon.com \
--to=dev-felipe.jensen@duagon.com \
--cc=dev-josejavier.rodriguez@duagon.com \
--cc=linux-kernel@vger.kernel.org \
--cc=morbidrsa@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®