The installer tries to start pcmciautils with the i82365 bridge module as a reasonable default. This worked fine in Edgy. Using Feisty's current alternate CD (in VMware), I get:
BUG: warning at lib/kref.c:32/kref_get()
[<c01044b2>] dump_trace+0x192/0x1c0
[<c01044f8>] show_trace_log_lvl+0x18/0x30
[<c0104b4f>] show_trace+0xf/0x20
[<c0104c15>] dump_stack+0x15/0x20
[<c01d2a5c>] kref_get+0x3c/0x40
[<c01d1e2f>] kobject_get+0xf/0x20
[<c0234e3e>] get_device+0xe/0x20
[<c02353cc>] device_add+0xc/0x4b0
[<c0239027>] platform_device_add+0xf7/0x150
[<ec98ed9f>] init_i82365+0x2f/0x47e [i82365]
[<c0132aaf>] sys_init_module+0x15f/0x1930
[<c0102f37>] syscall_call+0x7/0xb
[<b7f682be>] 0xb7f682be
=======================
BUG: unable to handle kernel NULL pointer dereference at virtual address 00000000
printing eip:
c02d3b63
*pde = 00000000
Oops: 0002 [#1]
Modules linked in: i82365 rsrc_nonstatic pcmcia_core ide_cd ide_disk ide_floppy ide_generic pdc202xx_old aec62xx alim15x3 amd74xx atiixp cmd64x cs5530 cs5535 cy82c693 hpt34x it821x ns87415 opti621 sc1200 serverworks siimage sis5513 trm290 usb_storage libusual fbcon tileblit font bitblit softcursor vga16fb vgastate vesafb usbserial usbkbd thermal processor fan sg sr_mod cdrom floppy generic piix psmouse uhci_hcd usbcore ata_piix libata scsi_mod evdev
CPU: 0
EIP: 0060:[<c02d3b63>] Not tainted VLI
EFLAGS: 00010002 (2.6.19-7-386 #2)
EIP is at __down+0x53/0xd6
eax: eaeb3df0 ebx: ec9a083c ecx: 00000000 edx: 00000000
esi: 00000286 edi: ea81b030 ebp: 3c9a0844 esp: eaeb3de4
ds: 007b es: 007b ss: 0068
Process modprobe (pid: 5957, ti=eaeb2000 task=ea81b0e0 task.ti=eaeb2000)
Stack: 00000001 ea81b0e0 c01152f0 ec9a0844 00000000 ec9a0768 ec9a083c ec9a0815
ec9a080c c02d3997 00000000 eaeb2000 c02374bb ec9a0768 ec9a0768 c0338100
c02366d6 ec9a0768 00000000 c023573c ec9a07d0 c02ebae5 ec9a080c ec9a0824
Call Trace:
[<c02d3997>] __down_failed+0x7/0x10
[<c02374bb>] device_attach+0x2b/0x80
[<c02366d6>] bus_attach_device+0x26/0x60
[<c023573c>] device_add+0x37c/0x4b0
[<c0239027>] platform_device_add+0xf7/0x150
[<ec98ed9f>] init_i82365+0x2f/0x47e [i82365]
[<c0132aaf>] sys_init_module+0x15f/0x1930
[<c0102f37>] syscall_call+0x7/0xb
[<b7f682be>] 0xb7f682be
=======================
Code: 00 89 7c 24 04 c7 44 24 08 f0 52 11 c0 c7 07 02 00 00 00 9c 5e fa 83 0c 24 01 8d 6b 08 8b 55 04 8d 44 24 0c 89 45 04 89 6c 24 0c <89> 02 8b 43 04 89 54 24 10 8d 50 01 89 53 04 01 03 0f 98 c0 84
EIP: [<c02d3b63>] __down+0x53/0xd6 SS:ESP 0068:eaeb3de4
I had to retype the oops report because I have no way that I can think of to copy it out of d-i any other way at that stage (this appears to confuse d-i sufficiently that it can't mount the CD). I hope it was worth it. Please investigate this urgently, as it's causing problems for Herd 1; however I may instigate a workaround of some kind.
Hmm, it looks like the CD mount failure is due to something else ...