ACK/Cmnt: [PATCH v2 0/1][SRU][OEM-5.6] Fix system boot hang at efi_tpm_eventlog_init

[Stefan Bader]

On 26.08.20 07:59, You-Sheng Yang wrote:
> BugLink: https://bugs.launchpad.net/bugs/1892827
> 
> [Impact]
> Some systems may hang at boot. With "earlycon=efifb", more debug
> messages for EFI are dumpped. Last call trace as:
> 
>   RIP: 0010:efi_tpm_eventlog_init+0x1aa/0x2a2
>   ...
>   Call Trace:
>    efi_config_parse_tables+0x12d/0x213
>    efi_init+0x218/0x36b
>    ? e820__print_table+0x65/0x6e
>    setup_arch+458/0xa84
> 
> This affects all mainline kernel since v5.2 to v5.7.6. Fix available in
> v5.8-rc3 and has been backported to v5.7.7, v5.4.50.
> 
> [Fix]
> Commit ea3cdcaa43b0 ("efi/tpm: Verify event log header before parsing")
> fixes commit 6b0326190205 ("efi: Attempt to get the TCG2 event log in
> the boot stub").
> 
> [Test Case]
> So far it can be easily reproduced with a green installation of Focal on
> USB key on a few Dell platforms.
> 
> [Regression Potential]
> Low. This checks validity of tmp event log header before parsing them,
> so it affects only platforms might generate invalid entries only.

And what would likely be the effect of either a valid event log entry being
accidentally rejected or an invalid one accidentally not detected (assuming the
latter is as before)?

> 
> [Other Info]
> This affects -generic for eoan and oem-5.6 for focal, but the former has
> be EOL-ed.
> 
> Fabian Vogt (1):
>   efi/tpm: Verify event log header before parsing
> 
>  include/linux/tpm_eventlog.h | 14 +++++++++++++-
>  1 file changed, 13 insertions(+), 1 deletion(-)
> 
Acked-by: Stefan Bader <stefan.bader at canonical.com>