geom_disk: Print attachment in boot log like newbus

[jrtc27]

Summary

If the attachment is known, print "GEOM: new disk A on B" so the
hierarchy is discernible from the log.

MFC after: 1 month

[mav]

I suppose 3 printf's may cause garbled output if other printf's are used by something.

[jrtc27]

In D46470#1059232, @mav wrote:

I suppose 3 printf's may cause garbled output if other printf's are used by something.

Yeah, I could use an sbuf, or duplicate the format string for the two cases. I don't really know what's best, so I picked the simplest of the three... depends how important this is really.

[imp]

In D46470#1059235, @jrtc27 wrote:

In D46470#1059232, @mav wrote:

I suppose 3 printf's may cause garbled output if other printf's are used by something.

Yeah, I could use an sbuf, or duplicate the format string for the two cases. I don't really know what's best, so I picked the simplest of the three... depends how important this is really.

Please use an sbuf. They will be garbled on systems with lots of disks because each target can proceed in parallel. This used to be a big problem with CAM until Scott fixed it.

[jrtc27]

In D46470#1059236, @imp wrote:

In D46470#1059235, @jrtc27 wrote:

In D46470#1059232, @mav wrote:

I suppose 3 printf's may cause garbled output if other printf's are used by something.

Yeah, I could use an sbuf, or duplicate the format string for the two cases. I don't really know what's best, so I picked the simplest of the three... depends how important this is really.

Please use an sbuf. They will be garbled on systems with lots of disks because each target can proceed in parallel. This used to be a big problem with CAM until Scott fixed it.

Ok.

[jrtc27]

In D46470#1059237, @jrtc27 wrote:

In D46470#1059236, @imp wrote:

In D46470#1059235, @jrtc27 wrote:

In D46470#1059232, @mav wrote:

I suppose 3 printf's may cause garbled output if other printf's are used by something.

Yeah, I could use an sbuf, or duplicate the format string for the two cases. I don't really know what's best, so I picked the simplest of the three... depends how important this is really.

Please use an sbuf. They will be garbled on systems with lots of disks because each target can proceed in parallel. This used to be a big problem with CAM until Scott fixed it.

Ok.

sbuf_new_auto? sbuf_new with NOWAIT? sbuf_new with a fixed buffer, and if so, how big? Or just do the stupid "hard-code each printf" that avoids any allocation and sbuf fanciness?

[jrtc27]

Avoid splitting printf via sneaky conditional operator