Re: More data on open-source firmware crash
On Monday 23 February 2009 07:30:13 Larry Finger wrote: Francesco Gringoli wrote: do you mind testing this firmware? It's not the solution, but can help us understanding if we should follow this way. Download at http://www.ing.unibs.it/~gringoli/fwtest.tar.gz Before using this firmware please recompile b43 changing these two definitions in b43.h #define B43_MARKER_ID_REG 52 #define B43_MARKER_LINE_REG 53 I coded the firmware so that it will raise a B43_DEBUGIRQ_MARKER with id 10, line 100 if the condition I'm thinking to is true. You will see (I hope) in dmesg. I ran the test firmware until there was a failure. The B43_DEBUGIRQ_MARKER was not present in the dmesg output. I also did as Michael suggested and preserved the depth of the tx_status queue, as well as the maximum depth observed. The latter value is 16; therefore, it is possible that the queue is being overrun. It is certainly full if the limit really is 16. I still need to look at the detailed data from the last run, but it is late here today. Note again that I expect the tx status queue is full check at the _start_ of the TX processing in firmware. Because if we started TX and later figure out the status queue is full, we can't really do anything about it. So we'd have to check that before even fetching the frame from the DMA. -- Greetings, Michael. ___ Bcm43xx-dev mailing list Bcm43xx-dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/bcm43xx-dev
Re: More data on open-source firmware crash
On Feb 22, 2009, at 8:10 PM, Larry Finger wrote: Francesco and Lorenzo, I modified my driver source to dump the firmware machine state whenever the b43_dma_handle_txstatus routine was called with an out-of-order cookie. With proprietary firmware, the test of a flood ping in one job and repeated tcpperf transmissions in a second ran for 10 hours without a single failure. With the open-source firmware it failed after about 2 hours. Below are the saved status data. Listed for each item are the cookie, the sequence number, and the skb length. The 0x84 length values come from the ping. All of the out-of-order items come from tcpperf - is it significant that they are from the longer set? Note that a number of cookie/sequence pairs are missing, namely: 2064/9C1, 2066/9C2, 2068/9C3, 206A/9C4, 206C/9C5, 2072/9C7, 2076/9C9, and 207A/9CB. Cookie 206E is missing, but the next sequence (9C6) was attached to cookie 2070. Larry, do you mind testing this firmware? It's not the solution, but can help us understanding if we should follow this way. Download at http://www.ing.unibs.it/~gringoli/fwtest.tar.gz Before using this firmware please recompile b43 changing these two definitions in b43.h #define B43_MARKER_ID_REG 52 #define B43_MARKER_LINE_REG 53 I coded the firmware so that it will raise a B43_DEBUGIRQ_MARKER with id 10, line 100 if the condition I'm thinking to is true. You will see (I hope) in dmesg. Thanks, bye -FG ___ Bcm43xx-dev mailing list Bcm43xx-dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/bcm43xx-dev
Re: More data on open-source firmware crash
Francesco Gringoli wrote: do you mind testing this firmware? It's not the solution, but can help us understanding if we should follow this way. Download at http://www.ing.unibs.it/~gringoli/fwtest.tar.gz Before using this firmware please recompile b43 changing these two definitions in b43.h #define B43_MARKER_ID_REG 52 #define B43_MARKER_LINE_REG 53 I coded the firmware so that it will raise a B43_DEBUGIRQ_MARKER with id 10, line 100 if the condition I'm thinking to is true. You will see (I hope) in dmesg. I ran the test firmware until there was a failure. The B43_DEBUGIRQ_MARKER was not present in the dmesg output. I also did as Michael suggested and preserved the depth of the tx_status queue, as well as the maximum depth observed. The latter value is 16; therefore, it is possible that the queue is being overrun. It is certainly full if the limit really is 16. I still need to look at the detailed data from the last run, but it is late here today. Larry ___ Bcm43xx-dev mailing list Bcm43xx-dev@lists.berlios.de https://lists.berlios.de/mailman/listinfo/bcm43xx-dev
