Bug 210449
Summary: | acpi_device_add: buffer overflow in strcpy | ||
---|---|---|---|
Product: | ACPI | Reporter: | Jethro Beekman (kernel) |
Component: | Config-Other | Assignee: | acpi_config-other |
Status: | RESOLVED CODE_FIX | ||
Severity: | normal | CC: | decui, erik.kaneda, rjw, Robert.Moore, rui.zhang, smfrench |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | next-20201201 5.11.0-rc2 | Subsystem: | |
Regression: | No | Bisected commit-id: | |
Attachments: |
ACPI tables dump
Oops on boot in acpi_device_add in 5.11-rc2 |
Description
Jethro Beekman
2020-12-02 10:16:17 UTC
Created attachment 293905 [details]
ACPI tables dump
Issue remains in Torvalds's current master branch I see there are some ACPICA changes in upstream kernel recently, but I'm not sure if they're included in the 23th Dec' git tree. Like the one below 32cf1a12cad43358e47dac8014379c2f33dfbed4 please check if the commit is included in your previous test. and better to confirm if the problem still exists in the latest upstream kernel or not. This is now in 5.11.0-rc2. Yes this includes the commit you mention. I am seeing a similar oops in 5.10-rc2 kernel (Ubuntu mainline build) Steve, is that a typo or are you actually testing 5.10-rc2 now (that's quite old)? (In reply to Steve French from comment #5) > I am seeing a similar oops in 5.10-rc2 kernel (Ubuntu mainline build) Sorry about the typo - "5.11-rc2" (installed today) (In reply to Jethro Beekman from comment #6) > Steve, is that a typo or are you actually testing 5.10-rc2 now (that's quite > old)? Yes - thanks for catching that. Was a typo. Installed current mainline Ubuntu kernel (5.11-rc2) build, on an Ubuntu 19.04 VM in Azure. So a common theme here appears to be Azure hypervisor. Created attachment 294573 [details]
Oops on boot in acpi_device_add in 5.11-rc2
Oops info attached. Booted in Azure VM from Ubuntu's weekly mainline kernel build (5.11-rc2) crashes. 5.10 booted fine,
This is my v1 fix: https://lkml.org/lkml/2020/12/18/514 Yesterday Rafael J. Wysock sugested that I should make a new fix using kstrdup_const(): https://lkml.org/lkml/2021/1/8/48 So I posted v2 later yesterday: https://lkml.org/lkml/2021/1/8/53 So far, no comment yet. Let me ping the ACPI maintainers now since we know more people are being affected. From "Rafael J. Wysocki", one of the ACPI maintainers: "I'm going to apply the v2 early next week." (see https://lkml.org/lkml/2021/1/9/218) Thanks Jethro for your testing! (https://lkml.org/lkml/2021/1/9/30) FYI: Now the fix is in the linux-pm tree's branch bleeding-edge: https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git/commit/?h=bleeding-edge&id=a58015d638cd4e4555297b04bec9b49028369075 Fixed in 5.11-rc4 |