1 2005-01-22 Andrew Lunn <andrew.lunn@ascom.ch>
3 * src/if_i82559.c Added string.h to remove compiler warnings.
5 2004-08-12 Jani Monoses <jani@iv.ro>
7 * src/if_i82559.c: Fix builing with lwip.
9 2004-01-15 Jeff Daly <jeffrey.daly@intel.com>
11 * src/if_i82559.c: support for large EEPROMs
13 2004-01-05 Gary Thomas <gary@mlbassoc.com>
15 * src/if_i82559.c: Names of FLASH config types changed.
17 2003-02-27 Mark Salter <msalter@redhat.com>
19 * src/if_i82559.c (eth_set_mac_address): Correct writing
20 of wrong address to EEPROM.
22 2003-02-22 Mark Salter <msalter@redhat.com>
24 * src/if_i82559.c: Fix endian issues with EEPROM writing.
25 Use read-modify-write to set EEPROM mac address. Support
26 writing to EEPROM the mac address obtained from
27 CYGHWR_DEVS_ETH_INTEL_I82559_GET_ESA.
29 2003-02-12 Mark Salter <msalter@redhat.com>
31 * src/if_i82559.c: Remove restriction preventing combination of
32 *_GET_ESA and *_HAS_EEPROM.
34 2003-01-03 Gary Thomas <gary@mlbassoc.com>
37 * cdl/intel_i82559_eth_drivers.cdl: Allow finer control over
38 debug (chatter) by making control booldata.
40 2002-12-17 Mark Salter <msalter@redhat.com>
42 * src/if_i82559.c (CYGHWR_DEVS_ETH_INTEL_I82559_ENDIAN_NEUTRAL_IO): New
43 flag to support systems where PCI IO operations are not affected by
46 2002-10-06 Andrew Lunn <andrew.lunn@ascom.ch>
48 * src/if_i82559.c (pci_init_find_82559s): Changed scope of
49 max_interrupt_handle since its needed in other places.
50 * src/if_i82559.c (i82559_stop): Corrected wrong name of variable
51 in a diag_printf function.
53 2002-07-24 Gary Thomas <gary@mlbassoc.com>
55 * src/if_i82559.c (pci_init_find_82559s):
56 Allow platform to define CYGHWR_DEVS_ETH_INTEL_I82559_USE_MEMORY
57 which forces use of memory instead of I/O space for device access.
59 2002-06-14 Gary Thomas <gary@chez-thomas.org>
62 Need to include <pkgconf/io_eth_drivers.h> for proper configuration
63 of stand-alone (polled) vs. system (interrupt driven) mode.
65 2002-04-09 Hugo Tyson <hmt@redhat.com>
67 * src/if_i82559.c (i82559_stop): Free up any pending transmissions
68 to prevent MBUF store leaks if the calling thread - the one that's
69 shutting down the interface - priority ends up higher than (or
70 same as) the network service threads.
72 2002-03-26 Jonathan Larmour <jlarmour@redhat.com>
74 * src/if_i82559.c (wait_for_cmd_done): Don't assert if RUC_ADDR_LOAD
75 command doesn't clear - on some chips it never does.
77 2002-03-18 Hugo Tyson <hmt@redhat.com>
79 * src/if_i82559.c (pci_init_find_82559s): Move allocation of
80 statistics pointer for each device to here, out of
81 i82559_start(). It used to leak PCI window store every time the
82 interface cycled down/up. [Case 107110]
84 2002-02-25 Mark Salter <msalter@redhat.com>
86 * src/if_i82559.c: Don't call PacketRxReady in standalone configuration
87 unless ready to accept a packet. Limit packet delivery to one packet at
88 a time in standalone configuration.
90 2002-02-20 Hugo Tyson <hmt@redhat.com>
92 * include/i82559_info.h: New fields to record/control multicast
93 all reception, oversized packet reception and promiscuous mode
96 * src/if_i82559.c: Initial support for multicast address reception
98 (i82559_restart): New routine to restart the device after fiddling
99 with its setup - a tail recursion from i82559_start().
100 (i82559_configure): More generic flags passed in and handled.
101 (i82559_set_multicast): Routine to pass a set of addresses into
102 the device; it works out the bitmap et al for us!
103 (i82559_ioctl): Handle new case arms ETH_DRV_SET_MC_LIST and
104 ETH_DRV_SET_MC_ALL if defined. Reset and restart the device,
105 diddling its status between.
107 2002-02-19 Mark Salter <msalter@redhat.com>
109 * src/if_i82559.c (pci_init_find_82559s): Don't use IRQ support if
110 CYGPKG_IO_ETH_DRIVERS_STAND_ALONE.
112 2002-01-28 Mark Salter <msalter@redhat.com>
114 * src/if_i82559.c (pci_init_find_82559s): Removed CYG_ASSERT which
115 assumed system PCI window exists.
117 2002-01-10 Jonathan Larmour <jlarmour@redhat.com>
119 * src/if_i82559.c (find_82559s_match_func): Add some more compatible PCI
122 2002-01-03 Mark Salter <msalter@redhat.com>
124 * src/if_i82559.c (pci_init_find_82559s): Remove check that
125 assumes driver owns entire PCI window.
127 * cdl/intel_i82559_eth_drivers.cdl: Add
128 CYGNUM_DEVS_ETH_INTEL_I82559_MAX_TX_DESCRIPTORS and
129 CYGNUM_DEVS_ETH_INTEL_I82559_MAX_RX_DESCRIPTORS.
131 2001-11-19 Hugo Tyson <hmt@redhat.com>
132 2001-11-19 Anssi Pulkkinen <Anssi.Pulkkinen@ascom.ch>
134 * src/if_i82559.c (TxMachine): The test to see if there is a new
135 tx to start must also say "go" if the tx queue is full. Normally,
136 if the txqueue is full, a tx will be occurring at all times - so
137 one will complete soon, so the tx queue full flag will soon be
138 cleared, and this condition will recover. I suspect a subtle race
139 which effectively means a new tx is queued and fills the queue in
140 between two tx's, so no TxDone() follows, causes the hang which we
141 get without this extra test under high load. [CASE 106686]
143 2001-09-20 Gary Thomas <gthomas@redhat.com>
145 * src/if_i82559.c: Add declaration of 'i82559_stop' to quiet warning.
147 2001-09-13 Hugo Tyson <hmt@redhat.com>
149 * cdl/intel_i82559_eth_drivers.cdl:
151 Remove the tickle thread from this driver. The functionality is
152 now in the network timeout functionality which handles the
153 "deliver" calls. So it's only called if the net is quiet.
155 2001-09-12 Hugo Tyson <hmt@redhat.com>
157 * src/if_i82559.c (TxDone): Defensive programming; zero the Tx key
158 in the global data before making the callback. This change
159 appeared useful in our forebear, the EBSA-specific driver.
161 2001-08-29 Hugo Tyson <hmt@redhat.com>
163 * cdl/intel_i82559_eth_drivers.cdl: New CDL options called
164 CYGPKG_DEVS_ETH_INTEL_I82559_TICKLE_THREAD, .._PRIORITY and
165 .._DELAY to control the thread. By default this is all disabled
166 because it is not needed for a real network application which
167 itself proactively tries to use the net - it's only needed when
168 the app is totally passive, so the driver otherwise does not get
171 * src/if_i82559.c (starti82559ticklethread): New code to start a
172 thread (!) to catch lost interrupts and thus restart jammed
173 hardware. It's only possible if KERNEL and NET - RedBoot is
174 unaffected, and it's not needed for RedBoot 'cos RedBoot polls.
175 (TxMachine): An assert about the CU idle status sometimes fired;
176 made the code more conservative wrt timing here.
177 (comments): more documentation on the possible external configury
178 by .inl file, described the two TIMEOUT macros that can be set,
179 and indeed must be to use the tickling thread meaningfully.
181 2001-08-22 Gary Thomas <gthomas@redhat.com>
184 printf() is no longer a part of RedBoot. Thus all programs
185 must use diag_printf() and related functions instead.
187 2001-06-22 Jesper Skov <jskov@redhat.com>
189 * src/if_i82559.c (i82559_init): Handle EEPROMS without CRC.
191 2001-05-16 Jonathan Larmour <jlarmour@redhat.com>
193 * src/if_i82559.c (udelay): Use virtual vector rather than assuming
194 hal_delay_us() exists.
196 2001-04-09 Hugo Tyson <hmt@redhat.com>
198 * src/if_i82559.c (pciwindow_mem_alloc): If asked, by
199 CYGHWR_DEVS_ETH_INTEL_I82559_PCIMEM_DISCONTIGUOUS, check for
200 breaks in the physical address of PCI window memory, and discard
201 any block with a break in it, then redo the alloc. This is for
202 targets where smaller SDRAM chips are fitted than the address
203 width, so they end up with gaps in (physical) memory.
205 2001-04-09 Hugo Tyson <hmt@redhat.com>
207 * src/if_i82559.c (i82559_init): Support hardware with only one
208 serial EEPROM attached to one device, but multiple devices. This
209 generally involves moving some decisions for eg. programming
210 EEPROM or not, optionally from compiletime to runtime. Improved
211 the diagnostics a bit also, and fixed some bugs.
212 (eth_set_mac_address): now takes an arg to say whether we want to
213 program the EEPROM, or are just setting up the chip.
214 (i82559_ioctl): This is the only place we set that arg to write
215 the EEPROM, startup never tries to write EEPROM from any external
216 source such as configured ESA or RedBoot params.
218 2001-04-09 Hugo Tyson <hmt@redhat.com>
220 * src/if_i82559.c (i82559_start): Warnings fixed; unused vars
221 following previous change.
223 2001-04-09 Hugo Tyson <hmt@redhat.com>
225 * src/if_i82559.c (i82559_start): Do not wait for command
226 completion after the initial DUMPSTATS operation. The code to
227 wait was bogus; it was looking at the structure in the wrong way.
228 In any case, there is no need to wait, the wait_for_cmd_done() in
229 any following activity is good enough.
231 2001-03-28 Hugo Tyson <hmt@redhat.com>
233 * src/if_i82559.c (update_statistics): Fix warning - only declare
234 op_i82559 if we use it - from previous change.
236 2001-03-23 Hugo Tyson <hmt@redhat.com>
238 * src/if_i82559.c (i82559_can_send, i82559_send): Count up and
239 down entries into these routines, and always say "no" to
240 can_send() if we are recursing, ie. within_send != 0. This is a
241 good idea because it limits stack usage. The recursion happens
242 because of the very neccessary polling of rx state within the tx
243 routines - which can deliver packets up into the stack, which can
244 trigger a response tx and so on... Also fixed a harmless bug
245 where, after polling, the p_i82559 current device state pointer
246 pointer to the wrong device, if we polled both devs regardless; it
247 only caused problems with the within_send recursion count.
249 * include/i82559_info.h (I82559): Add within_send field.
251 2001-03-15 Hugo Tyson <hmt@redhat.com>
253 * src/if_i82559.c (i82559_init): Support external macro
254 CYGHWR_DEVS_ETH_INTEL_I82559_GET_ESA(...), which if defined, is
255 used to acquire a MAC address, on the assumption that there is no
256 EEPROM present. CYGHWR_DEVS_ETH_INTEL_I82559_HAS_NO_EEPROM
257 confirms that there is no EEPROM attached to the 82559(s) so we
258 can omit all that code. It's in the negative sense because it's
261 2001-03-13 Hugo Tyson <hmt@redhat.com>
263 * src/if_i82559.c (CYGHWR_DEVS_ETH_INTEL_I82559_DEMUX_ALL): Fully
264 enable "DEMUX_ALL" option; all devices are scanned for
265 activity/wedging/interrupts at every opportunity. Multiple
266 devices on the same interrupts are now supported.
268 2001-03-12 Hugo Tyson <hmt@redhat.com>
270 * src/if_i82559.c (i82559_configure): New routine combines actions
271 of eth_set_promiscuous_mode() and eth_set_config() both of which
272 disappear. Thus the device is always configured correctly from
273 the start. [from a patch from Martin Buck <martin.buck@ascom.ch>
274 via the EBSA285 driver]
275 (eth_dsr): Do not call upper layer unless we are in a networked
276 [non-RedBoot, non-polled] environment. This is necessary because
277 we call our own DSR from foreground to unstick after a lost
279 (find_82559s_match_func): New routine to discover disparate
280 devices on the PCI bus. Used in pci_init_find_82559s().
281 (dump_txcb): Various junk debug functions removed.
283 2001-03-12 Hugo Tyson <hmt@redhat.com>
285 * src/if_i82559.c: Some tidying up, and two main additions to the
286 mechanisms used for managing this nasty entomological device.
287 (i82559_can_send): [amongst other places] if it is defined, use
288 CYGHWR_DEVS_ETH_INTEL_I82559_MISSED_INTERRUPT(p_i82559) to detect
289 that we missed an interrupt - and if so, call the ISR and DSR
290 directly. This is necessary for edge-triggered interrupt
291 controllers being fed by this level-sending device, where an
292 internal event in the device can keep the line low after the code
293 thinks that all events have been handled.
294 (Check82559TxLockupTimeout): This routine runs a timeout (if the
295 HAL provides a pair of watchdog-like macros) which performs a
296 selective reset of the device if it takes too long to transmit.
297 This happens rarely, but when it happens this is the only way out;
298 the tx machine is fixated on one tx and never comes out of it.
301 Sadly these two techniques to escape lockup only work if the
302 driver is called (ie. the stack tries to tx); incoming packets
303 cannot unwedge a device. We might need to add a poll from the
304 stack in future if this causes problems.
306 * include/i82559_info.h (I82559): Added two words for managing tx
307 lockup timeouts since this must be per-device.
309 2001-03-02 Hugo Tyson <hmt@redhat.com>
311 * src/if_i82559.c: Remove "Platform specific - defaults provided"
312 definitions for CYGHWR_INTEL_I82559_PCI_MEM_MAP_BASE/SIZE; all
313 platforms now provide these.
314 Added a little further disclaimer to the comment about LE/BE/GE.
316 2001-03-01 Hugo Tyson <hmt@redhat.com>
318 * src/if_i82559.c: Global change: what was HAL_READ_UINTxx or
319 HAL_WRITE_UINTxx are now READMEMxx or WRITEMEMxx - just
320 convenience macros that do all the volatile casts we want there.
321 Definitions of INL, INW, INB and OUTL, OUTW, OUTB recast in terms
322 of HAL_READ_UINTxx or HAL_WRITE_UINTxx so that they will use
323 proper IO operations on those CISCy mainframes that have a special
326 2001-03-01 Hugo Tyson <hmt@redhat.com>
328 * src/if_i82559.c: Much more configury added, so that I can use it
329 with a new target board. Including, but not limited to:
331 o An essay about addressing, big endian, little endian and GIB
332 endian (sic) and how we treat the various types of access within
333 this module. In other words, a lot of comments.
335 o CYGNUM_DEVS_ETH_INTEL_I82559_SEPARATE_MUX_INTERRUPT to clarify
336 that such an interrupt is SEPARATE from any simplex intrs that are
339 o CYGHWR_DEVS_ETH_INTEL_I82559_DEMUX_ALL for hardware where the
340 the only interrupt is a multiplexed (wire or'd) one. Plus
341 associated macros for direct manipulation of interrupt masks and
344 o Support for external definition from the platform HAL or driver
345 config module of CYGHWR_INTEL_I82559_PCI_MEM_MAP_BASE and SIZE.
346 Also CYGHWR_INTEL_I82559_PCI_VIRT_TO_BUS.
348 o LE definitions of a few macros that were missing.
350 o Separated definitions of structure offsets from BE/LE
351 definitions of consts within the words. Offsets vary with GIB
352 endian, for those CPUs who just flip bits. No change for BE
353 machines, needed for LE machines to work. To enable address-flip
354 within structures, define CYG_ADDRESSING_IS_GIBENDIAN.
356 o Re-org of mux/simplex deliver and isr's for new configuration.
358 o Couple of extra debug printouts.
360 2001-01-26 Jesper Skov <jskov@redhat.com>
362 * src/if_i82559.c: Added FIXME for promiscuous mode.
364 2001-01-25 Jesper Skov <jskov@redhat.com>
366 * src/if_i82559.c: Replace CYGNUM_HAL_INTERRUPT_PCI_IRQ with
367 CYGNUM_DEVS_ETH_INTEL_I82559_MUX_INTERRUPT which platform .inl
368 must provide when appropriate. Changed _deliver functions to match
370 * include/i82559_info.h: Remove debug hacks.
373 * src/if_i82559.c: Moved device descriptors into platform header.
374 Allow individual devices to hardwire ESA. Handle N devices instead
376 * include/i82559_info.h: ESA hardwired flag added.
377 * cdl/intel_i82559_eth_drivers.cdl: Device details moved to
380 * src/if_i82559.c (i82559_start): Poll status after dump command.
382 2001-01-24 Jesper Skov <jskov@redhat.com>
384 * src/if_i82559.c: [changes from yesterday] Some more endian
385 fiddle, and the device coughs out the initial ARP packets.
386 (wait_for_cmd_done): semantics changed to wait for pending CU
388 (i82559_reset): Cleaned up to spec. Init register bases after
390 Fix initialization of hardwired ESA.
391 Correct configuration command.
393 (eth_set_mac_address): Set driver's ESA as well. Proper check for
395 Fix LE structure offsets.
396 (i82559_start): Call new eth_set_config to enable device. Last
398 Remove a few printfs.
400 2001-01-23 Jesper Skov <jskov@redhat.com>
402 * src/if_i82559.c: Debug hackery and some endian issues resolved.
404 2001-01-22 Jesper Skov <jskov@redhat.com>
406 * src/if_i82559.c: Major changes, getting rid of device structures
407 in favor of HAL IO accessor macros. Also added endian conversion
408 macros where required.
409 * include/i82559_info.h: Change type of device structures to char.
411 2001-01-19 Jesper Skov <jskov@redhat.com>
413 * cdl/intel_i82559_eth_drivers.cdl: Hack for just one interface
415 * src/if_i82559.c: Fix IO functions. Handle EEPROM not being
418 * src/if_i82559.c: Use uncached/physical address translation
421 2001-01-15 Jesper Skov <jskov@redhat.com>
423 * src/if_i82559.c: Handle endian differences between controller
424 and CPU. Leave usdelay to HAL. Hack around PCI-base-at-0
426 Increase SK_DELAY (not that it changed anything, but now it's to
427 the spec), hardwire static ESA.
429 2001-01-12 Jesper Skov <jskov@redhat.com>
431 * src/if_i82559.c (pci_init_find_82559s): Check that device has
432 been found before accessing it.
433 (pci_init_find_82559s): Only delay if devices were
434 found. Recognize 82559ER code as well.
436 2001-01-10 Jesper Skov <jskov@redhat.com>
438 * src/if_i82559.c: Minor hacks to get it to build.
439 * include/i82559_info.h: Same.
440 * cdl/intel_i82559_eth_drivers.cdl: Same.
442 2000-12-07 Jesper Skov <jskov@redhat.com>
444 * cdl/intel_i82559_eth_drivers.cdl: Cloned from the EBSA driver.
445 * include/i82559_info.h: Same.
446 * src/if_i82559.c: Same.
448 2000-11-19 Gary Thomas <gthomas@redhat.com>
450 * src/if_ebsa285.c (pci_init_find_82559s): Intel has at least
451 two devices equivalent to the 82559. Support both (0x1229, 0x01030).
453 2000-10-05 Hugo Tyson <hmt@redhat.com>
455 * src/if_ebsa285.c: Deal with device interrupts in a nested
456 fashion - disable/restore is the semantics now, rather than
457 unconditionally unmasking. Also go directly to the 21285 PIC's
458 interrupt control registers to gain atomicity for these. Poll for
459 ready received packets when acknowledging an interrupt in the
460 tranmitting world; a race here could lose an Rx interrupt. Which
461 doesn't matter on a busy system, but in quieter times... there
462 will always be such a race because of the vague way the i82559's
463 status bits reflect how it's yanking the interrupt line; you have
464 to poll until the interrupt is gone before returning else spurious
465 interrupt failures occur. The issue is to close the window as
466 tightly as possible, which this change achieves at a minor cost in
467 performance - because of the time spent polling when not required.
469 2000-09-11 Hugo Tyson <hmt@cygnus.co.uk>
471 * src/if_ebsa285.c (i82559_poll): Only diddle the interface we
472 were asked to. This is more correct in terms of the intent of the
473 API, though it shouldn't really be necessary.
475 2000-09-06 Hugo Tyson <hmt@cygnus.co.uk>
477 * src/if_ebsa285.c (pci_init_find_82559s): Add asserts and an
478 unconditional check that the PCI window as configured matches the
479 address and size of the pci_window region from the MLT. This is
480 here because at present the MLT and CT cannot intercommunicate
481 enough. The separation of the PCI window is needed because
482 otherwise the malloc heap will eat all memory. [This is related
483 to CR 902624-CR, "MLT needs to be configuration aware"]
485 2000-09-01 Hugo Tyson <hmt@cygnus.co.uk>
487 * OVERVIEW: This is part of the change to the network stack to
488 greatly reduce latencies both of (other) DSRs and of thread
489 scheduling. All the work that the network stack *and* individual
490 ether drivers used to do in DSRs (including alarm callbacks and
491 data copies to/from the device memory) is moved into a "fast
492 network thread" instead. It calls a device's "deliver" function
493 to do the work that was previously in the DSR. This is a separate
494 thread so that it can be set higher priority than application
495 threads in order to minimize packet loss (depending on the
496 driver), if required (the application threads presumed to be
497 higher priority in turn than the network thread). A crucial
498 consequence of this is that we are no longer locking against DSRs,
499 so a plain mutex can be used rather than the global scheduler
500 lock, thus simplifying all the splfoo/splx() style functions.
502 * src/if_ebsa285.c: Minor: fix the big assert in i82559_send()
503 which suffered a race condition when called from the fast thread
504 rather than from a DSR. Major: Add a "deliver" entry to the
505 interface record for the "fast thread" implementation of the
506 network internal comms system. Provide a pass-up DSR to the
507 logical ether driver's DSR and appropriate delivery routine(s).
508 i82559_poll() now calls i82559_deliver() rather than the DSR. Add
509 valid data for mux'd DSR to pass on up.
511 2000-09-01 Hugo Tyson <hmt@cygnus.co.uk>
513 * tests/test_net_realtime.h: Tighten up the latency requirements
514 by a factor of 5; it all seems happy, so committed.
516 2000-08-25 Hugo Tyson <hmt@cygnus.co.uk>
518 * src/if_ebsa285.c (i82559_ioctl): A little further diddling; have
519 a bool to say whether the dot3 info is filled in.
521 2000-08-24 Hugo Tyson <hmt@cygnus.co.uk>
523 * cdl/ebsa285_eth_drivers.cdl: Do not export a symbol for the
524 device info file (include/ebsa285_info.h) since nobody needs
525 (portably) to include it now.
527 * src/if_ebsa285.c (i82559_ioctl): Handle new ioctl calls
528 ETH_DRV_GET_IF_STATS_UD and ETH_DRV_GET_IF_STATS to get loads of
529 statistical information out. _UD means update. The nonUD one can
530 be used instead mostly, if we find the performance hit too large.
531 This should allow SNMP (a) to not explode, (b) to get useful info
532 out of other device implementations than this one.
534 * include/ebsa285_info.h: Remove all the macro cruft for feature
535 detecting of lots of individual statistics; we now just have a
536 catch-all struct that SNMP uses, defined in the common ether
539 2000-08-15 Hugo Tyson <hmt@cygnus.co.uk>
541 * src/if_ebsa285.c (PacketRxReady): Put back the check for very
542 small packets into the driver; the layer above checks for that
543 (defensive programming) but only *after* asserting that the size
544 is large enough, to help detect that scenario from other drivers.
545 I believe we only have struct ether_header available if CYGPKG_NET
546 but I could be wrong.
549 2000-08-08 Hugo Tyson <hmt@cygnus.co.uk>
551 * src/if_ebsa285.c (eth_set_promiscuous_mode):
552 - ccs->config_bytes[18]=0x70;
553 + ccs->config_bytes[18]=0x72; // Keep the Padding Enable bit
554 ...otherwise short frame sends don't work in promisc mode.
557 2000-08-07 Gary Thomas <gthomas@redhat.com>
559 * src/if_ebsa285.c (pciwindow_mem_alloc): Take out very noisy debug.
561 2000-08-03 Gary Thomas <gthomas@redhat.com>
563 * src/if_ebsa285.c: Changes for stand-alone mode.
565 * cdl/ebsa285_eth_drivers.cdl: Ethernet driver package hierarchy changed.
566 Add option to control number of interfaces.
568 2000-07-28 Hugo Tyson <hmt@cygnus.co.uk>
570 * src/if_ebsa285.c (PacketRxReady): Do not attempt to forward
571 short packets; eth_drv.c assumes there is at least a header there.
572 (i82559_recv): Also be more careful and ASSERTive about -ve buffer
573 sizes; be more defensive about sglists. [CASE 104206]
575 2000-07-26 Gary Thomas <gthomas@redhat.com>
577 * src/if_ebsa285.c: Update for new eth_drv interfaces.
579 2000-07-18 Hugo Tyson <hmt@cygnus.co.uk>
581 * src/if_ebsa285.c (i82559_poll): Fill in the flesh of this, it
582 just calls ISR and DSR repeatedly.
583 (i82559_start): Look in the device record for promiscuous mode
584 flag; it should be passed though the common layer, but it's not
585 [yet] - this change from Andrew Lunn/ASCOM. Also a fix and delay
586 to the promisc mode code per se.
588 2000-07-17 Hugo Tyson <hmt@cygnus.co.uk>
590 * src/if_ebsa285.c (i82559_poll): New function, just to fill in
591 the interface record; not used.
593 2000-06-27 Hugo Tyson <hmt@cygnus.co.uk>
595 * cdl/ebsa285_eth_drivers.cdl: Add sesquipedalian option
596 CYGDBG_DEVS_ETH_ARM_EBSA285_KEEP_82559_STATISTICS in (now)
597 component CYGDBG_DEVS_ETH_ARM_EBSA285_KEEP_STATISTICS to control
598 keeping (well, harvesting really) the i82559's internal stats.
599 Reputedly, it doesn't service the net whilst this is happening, so
600 it could be viewed a bad thing. Hence the option.
602 * include/ebsa285_info.h: Only describe the I82559_COUNTERS
603 i82559_counters[2]; structs if full stats are to be kept.
605 * src/if_ebsa285.c (update_statistics): Only include this if full
606 stats are to be kept.
608 2000-06-27 Hugo Tyson <hmt@cygnus.co.uk>
610 * src/if_ebsa285.c (ResetRxRing): Re-do the management of the
611 RxRing; have an end-of-list flag (EL) in the last entry, and as
612 you unload filled slots, drag it round after you.
614 2000-06-14 Hugo Tyson <hmt@cygnus.co.uk>
616 * cdl/ebsa285_eth_drivers.cdl: Add option to control statistics
619 * include/ebsa285_info.h: Act on it.
621 2000-06-13 Hugo Tyson <hmt@cygnus.co.uk>
623 * cdl/ebsa285_eth_drivers.cdl: Properly include the new header
624 file and define symbols to allow client code to get at it without
625 knowing the particular device driver name.
627 * include/ebsa285_info.h: New file: export various statistics
628 information about the driver for use by monitoring and
629 network-management systems. This requires exposing the
630 (otherwise) internal structures of the driver.
632 * src/if_ebsa285.c: remove a lot of structure definitions &c that
633 are now necessarily in the new header; add a couple of new
634 routines which provide status and update statistics from the
635 device into memory; tidy up control of whether stats-keeping is
638 2000-06-06 Hugo Tyson <hmt@cygnus.co.uk>
640 * cdl/ebsa285_eth_drivers.cdl (define_proc): Add #define of
641 CYGDAT_DEVS_ETH_DESCRIPTION in the config file for information.
643 2000-05-12 Hugo Tyson <hmt@cygnus.co.uk>
645 * tests/test_net_realtime.h (tnr_print_activity): New routine to
646 check the system is working, tidied up the API. It works!
648 2000-05-11 Hugo Tyson <hmt@cygnus.co.uk>
650 * cdl/ebsa285_eth_drivers.cdl: Added export of the test header
651 below, and config opts for controlling EEPROM writing and all the
652 status chatter as the device starts up.
654 * src/if_ebsa285.c: Reworked the code for reading and setting the
655 EEPROM that holds the MAC address. This is very ugly, but now
656 more reliable. Also tidied up printing cruft with neater
657 configury, and made it an option (for safety) whether it's
658 possible to write the EEPROM at all.
660 * tests/test_net_realtime.h: New file - it is intended to be used
661 by networking tests to verify that latency is not compromised by
662 the stack and driver. It's very platform specific, hence the
663 location in here. This is a preliminary version only.
665 2000-04-27 Hugo Tyson <hmt@cygnus.co.uk>
667 * src/if_ebsa285.c: A serious re-write. This cuts out a lot of
668 code from the old version and improves the performance greatly.
670 The cruft was mainly doing lots of explicit event communication
671 between the ISR and DSR, when in fact all the state needed is
672 present in the tx/rx rings. So both ISRs and DSRs regard their
673 call as an opportunity to progress everything they can, rather
674 than only dealing with one interrupt cause at a time; the
675 connection between them is now rather looser.
677 Interrups can now be re-enabled after the ISR (in other words they
678 are not masked in the ISR), no need to wait for the DSR, but in
679 consequence some DSR code must mask/unmask intrs as it works.
681 The 82559 appears to be a little slow in reacting to commands and
682 state changes, so some interrupts were being lost - or persisting
683 beyond their desired life - so there's some kinda polling code to
684 deal with that also. We also rely on the foreground to kind of
685 poll in the same way, in the send/can_send calls - we know the
686 stack will re-try if necessary, though this is rare.
688 The driver now works (albeit at much reduced performance) with as
689 few as 6 rx and tx buffers - in other words the "queue full/out of
690 rx buffers" states have been tested and all is well. It works
691 generally fine with 8 buffers of each kind.
693 The mux ISR and DSR are now rather more polled than the old
694 versions; we just try to do things with both devices (if active)
695 by simply calling each unitary ISR/DSR respectively.
697 I also re-ordered some of the code, moving utilities to the end of
698 the file and grouping together Tx and Rx machines a bit better.
700 2000-04-13 Hugo Tyson <hmt@cygnus.co.uk>
702 * src/if_ebsa285.c: Attribution to Ron Spence, Pacific Softworks
703 added as a contributor.
705 2000-04-07 Hugo Tyson <hmt@cygnus.co.uk>
707 * ecos.db: Re-organize device packages. This is a massive change
708 involving deleting all the sources for serial and ethernet drivers
709 from where they used to live in
710 packages/io/serial/current/src/ARCH/PLATFORM.[ch]
711 packages/net/drivers/eth/PLATFORM/current/src/...
712 and reinstating them in
713 packages/devs/serial/ARCH/PLATFORM/current/src/...
714 packages/devs/eth/ARCH/PLATFORM/current/src/...
716 All these new packages are properly defined in ecos.db, and are
717 all of type "hardware" so that a "target" can grab them.
719 This directory layout is descriptive of the devices we have right
720 now, arch and platform are separate levels just to make it easier
721 to navigate in the filesystem and similar to the HAL structure in
724 It is *not* prescriptive of future work; for example, the mythical
725 common highly-portable 16550 serial driver which works on many
726 targets would be called "devs/serial/s16550/current", or a serial
727 device for a particular board (cogent springs to mind) that can
728 work with different CPUs fitted is "devs/serial/cogent/current".
730 Changelogs have been preserved and replicated over all the new
731 packages, so that no history is lost.
733 The contents of individual source files are unchanged; they build
734 in just the same emvironment except for a very few cases where the
735 config file name changed in this movement.
737 Targets in ecos.db have been redefined to bring in all relevant
738 hardware packages including net and serial drivers (but the newly
739 included packages are only active if their desired parent is
742 The names of CDL options (and their #defines of course) stay the
743 same for the serial drivers, for backward compatibility.
745 * templates/*/current.ect: these have had CYGPKG_IO_SERIAL added
746 rather than it being in (almost) all target definitions.
748 2000-03-29 Hugo Tyson <hmt@cygnus.co.uk>
750 * src/if_ebsa285.c (i82559_recv): Be happy with NULLs in the
751 SGlist; it means the caller is out of memory so drop the packet on
752 the floor. Also remove a completely redundant test.
754 2000-03-06 Hugo Tyson <hmt@cygnus.co.uk>
756 * src/if_ebsa285.c (i82559_can_send): Update net driver to new
757 interface style. This is incomplete wrt promiscuous mode, but
758 that's probably about all.
760 2000-02-14 Gary Thomas <gthomas@cygnus.co.uk>
762 * cdl/ebsa285_eth_drivers.cdl: Update CDL to indicate multiple
765 2000-02-14 Hugo Tyson <hmt@cygnus.co.uk>
767 * src/if_ebsa285.c (pci_init_find_82559s): Tidy comments somewhat
768 and set debug and stats collecting defines to most friendly
771 2000-02-10 Hugo Tyson <hmt@cygnus.co.uk>
773 * src/if_ebsa285.c (PacketRxReady): Fix bug; current descriptor
774 was not being write back for the callback to use. Hence asserts
775 on state of rfd were firing in busy times - that leading rfd had
776 already been drained.
778 Also rationalized meaning of DEBUG printy symbols a bit - it's now
779 chatty during startup/foreground manipulations but not in any
780 performance related activities ie. rx or tx.
782 2000-02-09 John Dallaway <jld@cygnus.co.uk>
784 * cdl/ebsa285_eth_drivers.cdl:
786 Reparent under CYGPKG_NET_ETH_DRIVERS and tidy display strings.
788 2000-02-08 Hugo Tyson <hmt@cygnus.co.uk>
790 * src/if_ebsa285.c: New File.
791 * cdl/ebsa285_eth_drivers.cdl: New File.
793 Initial Checkin of EBSA285 Ethernet driver.
795 It's one monolithic file at present, and should be split up into a
796 more generic Intel 82559 driver plus platform-specific parts (PCI
797 et al) plus eCos/Red-Hat-BSD-stack parts.
799 //===========================================================================
800 //####ECOSGPLCOPYRIGHTBEGIN####
801 // -------------------------------------------
802 // This file is part of eCos, the Embedded Configurable Operating System.
803 // Copyright (C) 1998, 1999, 2000, 2001, 2002 Red Hat, Inc.
805 // eCos is free software; you can redistribute it and/or modify it under
806 // the terms of the GNU General Public License as published by the Free
807 // Software Foundation; either version 2 or (at your option) any later version.
809 // eCos is distributed in the hope that it will be useful, but WITHOUT ANY
810 // WARRANTY; without even the implied warranty of MERCHANTABILITY or
811 // FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
814 // You should have received a copy of the GNU General Public License along
815 // with eCos; if not, write to the Free Software Foundation, Inc.,
816 // 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
818 // As a special exception, if other files instantiate templates or use macros
819 // or inline functions from this file, or you compile this file and link it
820 // with other works to produce a work based on this file, this file does not
821 // by itself cause the resulting work to be covered by the GNU General Public
822 // License. However the source code for this file must still be made available
823 // in accordance with section (3) of the GNU General Public License.
825 // This exception does not invalidate any other reasons why a work based on
826 // this file might be covered by the GNU General Public License.
828 // Alternative licenses for eCos may be arranged by contacting Red Hat, Inc.
829 // at http://sources.redhat.com/ecos/ecos-license/
830 // -------------------------------------------
831 //####ECOSGPLCOPYRIGHTEND####
832 //===========================================================================