With kernel 2.6.18-rc1 I am experiencing strange delays before printk messages from my driver appear in the syslog. The last couple of printk lines only appear if I hit a key on the system keyboard. Even just pressing and releasing a shift key will do. Other input, like connecting with ssh from another computer and typing into that session or moving or clicking the mouse, doesn't help. At the same time, userspace syslog messages like those from sshd appear just fine. Example excerpt from /var/log/messages: > Jul 7 00:16:08 gx110 kernel: [ 746.583136] gigaset: 0: if_write() > Jul 7 00:16:08 gx110 kernel: [ 746.583154] gigaset: CMD Transmit (10 bytes): > Jul 7 00:16:08 gx110 kernel: [ 746.583203] bas_gigaset: setting ATREADY time > Jul 7 00:16:08 gx110 kernel: [ 746.585887] bas_gigaset: write_command: sent > Jul 7 00:16:08 gx110 kernel: [ 746.612869] gigaset: received response (6 byt > Jul 7 00:16:28 gx110 sshd[6134]: Accepted publickey for ts from 192.168.59.12 > Jul 7 00:28:26 gx110 kernel: [ 746.687844] gigaset: 0: if_write() > Jul 7 00:28:26 gx110 kernel: [ 746.687944] gigaset: CMD Transmit (17 bytes): > Jul 7 00:28:26 gx110 kernel: [ 746.688002] bas_gigaset: setting ATREADY time > Jul 7 00:28:26 gx110 kernel: [ 746.690803] bas_gigaset: write_command: sent Note how the PRINTK_TIME timestamps increase in sub-second increments while syslogd's timestamps show an enormous gap after 00:16:08. With kernel 2.6.17.1 on the same machine there are no such delays. All printk messages appear immediately, and the deltas in PRINTK_TIME and syslogd timestamps are quite consistent. Regards Tilman -- Tilman Schmidt E-Mail: [email protected] Bonn, Germany It is well known that a vital ingredient of success is not knowing that what you're attempting can't be done. [Terry Pratchett, Equal Rites]
Attachment:
signature.asc
Description: OpenPGP digital signature
- Follow-Ups:
- Re: Linux v2.6.18-rc1: printk delays
- From: john stultz <[email protected]>
- Re: Linux v2.6.18-rc1: printk delays
- Prev by Date: Re: [perfmon] Re: cpuinfo_x86 and apicid
- Next by Date: Implement class_device_update_dev() function
- Previous by thread: Re: [PATCH 38 of 39] IB/ipath - More changes to support InfiniPath on PowerPC 970 systems
- Next by thread: Re: Linux v2.6.18-rc1: printk delays
- Index(es):