From: Bjorn Helgaas <helgaas@kernel.org>
To: lirongqing <lirongqing@baidu.com>
Cc: "Bjorn Helgaas" <bhelgaas@google.com>,
"Philipp Stanner" <pstanner@redhat.com>,
"Krzysztof Wilczyński" <kwilczynski@kernel.org>,
linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] PCI: Fix pcim_iomap_regions() error path rollback and leak
Date: Mon, 9 Nov 2026 17:33:37 -0600 [thread overview]
Message-ID: <20261109233337.GA347778@bhelgaas> (raw)
In-Reply-To: <20261109113136.2295-1-lirongqing@baidu.com>
On Mon, Nov 09, 2026 at 07:31:36PM +0800, lirongqing wrote:
> From: Li RongQing <lirongqing@baidu.com>
>
> pcim_iomap_regions() has two problems in its error path.
>
> First, the rollback loop walks every bar index below the failing one
> and unmaps it, regardless of whether that bar is in the mask this call
> was asked to map. pcim_iounmap_region() matches devres entries by
> (type, bar) only, so it can release a mapping that an earlier
> pcim_iomap_region()/pcim_iomap_regions() call made. The driver is then
> left with an __iomem pointer that has been unmapped and a legacy table
> entry cleared. Restrict the rollback to bars in mask.
>
> Second, when pcim_add_mapping_to_legacy_table() fails, the current
> bar's mapping is not released: the rollback loop starts at --bar, so
> the failing bar itself is skipped, and the mapping stays registered
> until the driver detaches. Release it on this path as well.
>
> Fixes: bbaff68bf4a4 ("PCI: Add managed partial-BAR request and map infrastructure")
> Signed-off-by: Li RongQing <lirongqing@baidu.com>
Hoping for a Reviewed-by from Philipp.
If there's a v2, s/bar/BAR/ above to match other usage. If no v2,
I'll tweak that when applying.
> ---
> drivers/pci/devres.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/pci/devres.c b/drivers/pci/devres.c
> index f075e78..9bdb4de 100644
> --- a/drivers/pci/devres.c
> +++ b/drivers/pci/devres.c
> @@ -689,14 +689,18 @@ int pcim_iomap_regions(struct pci_dev *pdev, int mask, const char *name)
> goto err;
> }
> ret = pcim_add_mapping_to_legacy_table(pdev, mapping, bar);
> - if (ret != 0)
> + if (ret != 0) {
> + pcim_iounmap_region(pdev, bar);
> goto err;
> + }
> }
>
> return 0;
>
> err:
> while (--bar >= 0) {
> + if (!mask_contains_bar(mask, bar))
> + continue;
> pcim_iounmap_region(pdev, bar);
> pcim_remove_bar_from_legacy_table(pdev, bar);
> }
> --
> 2.9.4
>
prev parent reply other threads:[~2026-11-09 23:33 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-09 11:31 [PATCH] PCI: Fix pcim_iomap_regions() error path rollback and leak lirongqing
2026-11-09 11:35 ` sashiko-bot
2026-11-09 23:33 ` Bjorn Helgaas [this message]
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=20261109233337.GA347778@bhelgaas \
--to=helgaas@kernel.org \
--cc=bhelgaas@google.com \
--cc=kwilczynski@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=lirongqing@baidu.com \
--cc=pstanner@redhat.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.