From: Li Ming <ming4.li@intel.com>
To: linux-cxl@vger.kernel.org, rrichter@amd.com,
terry.bowman@amd.com, dan.j.williams@intel.com,
alison.schofield@intel.com, pengfei.xu@intel.com
Cc: Li Ming <ming4.li@intel.com>
Subject: [PATCH v2 0/2] Fix get a wrong pci host bridge in cxl_setup_parent_dport()
Date: Fri, 9 Aug 2024 08:27:48 +0000 [thread overview]
Message-ID: <20240809082750.3015641-1-ming4.li@intel.com> (raw)
The cxl_test unit test environment on qemu always hit below call trace
with KASAN enabled:
BUG: KASAN: slab-out-of-bounds in cxl_setup_parent_dport+0x480/0x530 [cxl_core]
Read of size 1 at addr ff110000676014f8 by task (udev-worker)/676[ 24.424403] CPU: 2 PID: 676 Comm: (udev-worker) Tainted: G O N 6.10.0-qemucxl #1
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS edk2-20240214-2.el9 02/14/2024
Call Trace:
<TASK>
dump_stack_lvl+0xea/0x150
print_report+0xce/0x610
? kasan_complete_mode_report_info+0x40/0x200
kasan_report+0xcc/0x110
__asan_report_load1_noabort+0x18/0x20
cxl_setup_parent_dport+0x480/0x530 [cxl_core]
cxl_mem_probe+0x49b/0xaa0 [cxl_mem]
The root cause is that a wrong host bridge was gotten from
dport->dport_dev in cxl_setup_parent_dport(). In
cxl_setup_parent_dport(), it always calls
to_pci_host_bridge(dport->dport_dev) to get a pci_host_bridge structure.
There are two issues in the implementation:
* to_pci_host_bridge(dport->dport_dev) should be used only for RCH
cases, dport->dport_dev points to a pci device rather than a pci host
bridge in VH cases. The solution is checking if dport is in RCH mode
then calling to_pci_host_bridge().
(Patch 1)
* In cxl_test unit test environment, cxl_test will create a emulated CXL
topology with emulated dports, the dport_dev of a emulated dport
points to a platform device. to_pci_host_bridge(dport->dport_dev) also
gets a wrong pci host bridge in the case. The solution is implementing
a new wrap function called __wrap_cxl_setup_parent_dport() on cxl_test
side, the function will filter all emulated dports, make sure only
real dports can be handled by cxl_setup_parent_dport().
(Patch 2)
v1 link: https://lore.kernel.org/linux-cxl/ZrHR+0w3bwM1Ik8h@xpf.sh.intel.com/T/#med6200e54ec12f09fdcc04571516adda261c9561
v2:
- Add call trace log into changelog
- Remove 'dev_is_platform(dport->dport_dev)' checking out of cxl driver
scope. Check if dport is emulated in cxl_test.
Li Ming (2):
cxl/pci: Get AER capability address from RCRB only for RCH dport
cxl/test: Skip cxl_setup_parent_dport() for emulated dports
drivers/cxl/core/pci.c | 10 ++++++----
tools/testing/cxl/Kbuild | 1 +
tools/testing/cxl/test/mock.c | 12 ++++++++++++
3 files changed, 19 insertions(+), 4 deletions(-)
--
2.40.1
next reply other threads:[~2024-08-09 8:58 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-08-09 8:27 Li Ming [this message]
2024-08-09 8:27 ` [PATCH v2 1/2] cxl/pci: Get AER capability address from RCRB only for RCH dport Li Ming
2024-08-09 8:27 ` [PATCH v2 2/2] cxl/test: Skip cxl_setup_parent_dport() for emulated dports Li Ming
2024-08-10 2:36 ` Pengfei Xu
2024-08-09 21:36 ` [PATCH v2 0/2] Fix get a wrong pci host bridge in cxl_setup_parent_dport() Dan Williams
2024-08-09 21:44 ` Ira Weiny
2024-08-12 6:49 ` Li, Ming4
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=20240809082750.3015641-1-ming4.li@intel.com \
--to=ming4.li@intel.com \
--cc=alison.schofield@intel.com \
--cc=dan.j.williams@intel.com \
--cc=linux-cxl@vger.kernel.org \
--cc=pengfei.xu@intel.com \
--cc=rrichter@amd.com \
--cc=terry.bowman@amd.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