Re: [PATCH v2 3/4] reftable/stack: move list lock to `struct reftable_stack`
- From
Karthik Nayak <karthik.188@gmail.com>
- Date
- Aug 26, 2026, 09:46 UTC
- Message-ID
- <CAOLa=ZSbOFmsNUfhsEESsD5rBqUq0iX=Ls_TgfRkeo5FQpamtQ@mail.gmail.com>
- In-Reply-To
- <ao1uqpCxFHlOyTV-@pks.im>
Patrick Steinhardt <ps@pks.im> writes:
Show 32 quoted lines
> On Mon, Aug 24, 2026 at 11:31:01AM +0200, Karthik Nayak wrote:
>> diff --git a/reftable/stack.c b/reftable/stack.c
>> index 703548417c..c3d4deff29 100644
>> --- a/reftable/stack.c
>> +++ b/reftable/stack.c
>> @@ -628,10 +630,16 @@ int reftable_stack_reload(struct reftable_stack *st)
>> }
>>
>> struct reftable_addition {
>> - struct reftable_flock tables_list_lock;
>> struct reftable_stack *stack;
>> struct reftable_write_options opts;
>>
>> + /*
>> + * While the list lock is acquired on the stack, we need to distinguish
>> + * which 'reftable_addition' is responsible for the lock. This avoids
>> + * clearing the lock of another 'reftable_addition'.
>> + */
>> + unsigned int locked : 1;
>> +
>> char **new_tables;
>> size_t new_tables_len, new_tables_cap;
>> uint64_t next_update_index;
>
> This feels somewhat fragile, as tracking state via a bit is very easy to
> get wrong. I would have preferred a solution where we didn't have to
> have this field and instead a more direct proxy. But there's not that
> many spots where we have to do this, so this may be okay-ish.
>
> Other than that this series looks good to me, thanks!
>
> PatrickI agree, I couldn't find another solution. The alternative being, we move the bitfield to the 'reftable_stack', but that is more fragile since there are more touch fields. So I think this is okay too.
Thanks