On 06/09/2015 06:55 AM, Richard Weinberger wrote:
Am 08.06.2015 um 12:07 schrieb Dongsheng Yang:- ubifs_assert(mutex_is_locked(&ui->ui_mutex)); if (!ui->dirty) { + if (!locked) { + /* + * It's a little tricky here, there is only one + * possible user of ubifs_dirty_inode did not do + * a budget for this inode. At the same time, this + * user is not holding the ui->ui_mutex. Then if + * we found ui->ui_mutex is not locked, we can say: + * we need to do a budget in ubifs_dirty_inode here. + */ + struct ubifs_budget_req req = { .dirtied_ino = 1, + .dirtied_ino_d = ALIGN(ui->data_len, 8) }; + + ret = ubifs_budget_space(c, &req); + if (ret) + goto out; + }So, this is the new case when ->dirty_inode() is called via generic_update_time()? Did you research whether you can detect that case also by looking at the flags parameter? I'd give I_DIRTY_TIME a try. This way you could get at least rid of the mutex_is_locked() usage.
Okey, after a reading, I'm afraid I can not think a better idea out. The flags between *old* cases and the *new* case can possiblly be same. Then we can't use the flags to filter the new case from old cases. But I think I can append a patch to add a support for lazytime here: if (flags == I_DIRTY_TIME) return; Thanx Yang
Thanks, //richard .
-- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
