[PATCH v2 3/3] tpm: allocate tpm event log if missing

[Ludwig Nussel]

On 6/11/26 11:57, Ilias Apalodimas wrote:
> [...]
> On Wed, 10 Jun 2026 at 13:52, Ludwig Nussel <ludwig.nussel at siemens.com> wrote:
> [...]
>>          if (IS_ENABLED(CONFIG_MEASURED_BOOT)) {
>> -               struct tcg2_event_log elog;
>>                  struct udevice *dev;
>>                  void *initrd_buf;
>>                  void *image_buf;
>>                  const char *s;
>>                  u32 rd_len;
>> -               bool ign;
>>
>> -               elog.log_size = 0;
>> -               ign = IS_ENABLED(CONFIG_MEASURE_IGNORE_LOG);
>> -               ret = tcg2_measurement_init(&dev, &elog, ign);
>> -               if (ret)
>> +               ret = tcg2_measurement_init(&dev, NULL);
> 
> I'd like the eventlog to be initialized before calling this (look below).

What if I removed the log parameter from those tcg2_* functions instead? 
The event log belongs to the tpm device IMO, so should be initialized 
with the device.
I am not sure what the intention of the current api is tbh. The event 
log basically documents when and how PCR registers were extended. We 
have to build that log ourselves as the TPM doesn't do it for us. The 
api weirdly allows to have partial local logs even though there's only 
one set of PCR registers.

>> [...]
>> index d41228f75a9..2270b038cf5 100644
>> --- a/lib/tpm_tcg2.c
>> +++ b/lib/tpm_tcg2.c
>> @@ -21,6 +21,7 @@
>>   #include <linux/unaligned/le_byteshift.h>
>>   #include "tpm-utils.h"
>>   #include <bloblist.h>
>> +#include <dm/devres.h>
>>
>>   int tcg2_get_pcr_info(struct udevice *dev, u32 *supported_bank, u32 *active_bank,
>>                        u32 *bank_num)
>> @@ -212,6 +213,9 @@ static int tcg2_log_append_check(struct tcg2_event_log *elog, u32 pcr_index,
>>          u32 event_size;
>>          u8 *log;
>>
>> +       if (!elog || !elog->log_size)
>> +               return 0;
>> +
>>          event_size = size + tcg2_event_get_size(digest_list);
>>          if (elog->log_position + event_size > elog->log_size) {
>>                  printf("%s: log too large: %u + %u > %u\n", __func__,
>> @@ -553,6 +557,9 @@ int tcg2_measure_data(struct udevice *dev, struct tcg2_event_log *elog,
>>          if (rc)
>>                  return rc;
>>
>> +       if (!elog)
>> +               elog = tcg2_platform_get_dev_log(dev);
>> +
> 
> Initializing this here feels wrong. This is a function that just
> measures and logs data. Why can't we fill in the eventlog address
> early before calling this? If we do so some of the checks above are
> also not needed

tcg2_platform_get_dev_log() is just a wrapper to access the private 
property of dev that keeps the event log address.
How about moving the call into tcg2_log_append_check() instead?

cu
Ludwig

-- 
Ludwig Nussel
Siemens AG
www.siemens.com