Lately I'm having some troubles with Netscape. The bug is very specific but I still couldn't fully trace its origin. Note that this bug never happened before and I'm using netscape 4.75 over several X builds. Netscape Mail client trashes its mailbox indexes located on ~/nsmail/.* (* for Index, Trash...). How and when it does that is still a mistery. The phenomena happens randomly and with any kind of mail in any of the mailboxes I have. It happens even on boxes that have been fully cleaned recently. Also, Netscape seems to crash randomly when reading these broken mailboxes: on reading a letter, on deleting, on downloading mail. In one of the events I could read every mail in the Inbox but when I deleted a letter Netscape crashed. The crash is even reflected on the kernel: kernel BUG at ll_rw_blk.c:711! invalid operand: 0000 CPU: 0 EIP: 0010:[__make_request+161/1488] EFLAGS: 00010282 eax: 0000001f ebx: c689c300 ecx: 00000005 edx: 00000029 esi: c689c300 edi: c0335bc0 ebp: 00000001 esp: c461bea8 ds: 0018 es: 0018 ss: 0018 Process netscape-commun (pid: 801, stackpage=c461b000) Stack: c026092b c0260c22 000002c7 c689c300 00000001 0000000c 00000000 001e8480 c0335bd8 c0335bd0 00000000 00000002 00000000 00000000 c017dc7f 000000fe c017e8bd c0335bc0 00000001 c689c300 c689c300 00000000 00000001 c461bf38 Call Trace: [devfsd_buf_size+56899/73016] [devfsd_buf_size+57658/73016] [blk_get_queue+63/84] [generic_make_request+257/272] [ll_rw_block+341/452] [writeout_one_page+52/76] [do_buffer_fdatasync+93/160] [generic_buffer_fdatasync+29/56] [writeout_one_page+0/76] [ext2_sync_file+89/260] [sys_write+139/160] [sys_fsync+74/104] [system_call+51/56] Code: 0f 0b 83 c4 0c 0f b6 46 15 0f b7 4e 14 8b 14 85 40 a2 32 c0 After this the index file is unreadable until reboot (mc hangs reading it). The system seems to continue normal operations but sync or some other disk operations may hang the machine. Netscape hangs if one tries to reload, even if we don't launch the mail client. Besides, doing this, it may risk a full freeze. Reboot has to be done either through Magic Key or through "AnyKey" (reset :) ). After boot my network card gets schizo - the kernel starts an avalanche of timeout warnings (It's a IBM EtherExpress with WakeOnLAN). All files in ~/nsmail are readable. There are no physical problems with the disk that could trigger such situation. The situation can only be solved by deleting the mail indexes. Ektanoor
