Hi,
The Qdisc drops does not seem to be reported to user space for the
default queuing discipline -- pfifo_fast_ops. I ran isic tests on a up
to test if the drops happen at all, and yes there were significant no of
packets dropped under heavy tx load. Here is a patch which reports just
the qdisc drops to the /proc/net/dev entry. Does this sound like a
good idea?
Comments welcome.
-Kiran
diff -X dontdiff -ruN linux-2.5.31/include/asm-i386/mmu.h statctr-2.5.31/include/asm-i386/mmu.h
--- linux-2.5.31/include/asm-i386/mmu.h Sun Aug 11 07:11:42 2002
+++ statctr-2.5.31/include/asm-i386/mmu.h Wed Aug 21 12:40:35 2002
@@ -1,6 +1,8 @@
#ifndef __i386_MMU_H
#define __i386_MMU_H
+#include <asm/semaphore.h>
+
/*
* The i386 doesn't have a mmu context, but
* we put the segment information here.
diff -X dontdiff -ruN linux-2.5.31/net/core/dev.c statctr-2.5.31/net/core/dev.c
--- linux-2.5.31/net/core/dev.c Sun Aug 11 07:11:30 2002
+++ statctr-2.5.31/net/core/dev.c Wed Aug 21 17:57:24 2002
@@ -1746,10 +1746,12 @@
NULL;
int size;
+ struct Qdisc *q = dev->qdisc;
+
if (stats)
size = sprintf(buffer, "%6s:%8lu %7lu %4lu %4lu %4lu %5lu "
"%10lu %9lu %8lu %7lu %4lu %4lu %4lu "
- "%5lu %7lu %10lu\n",
+ "%5lu %7lu %10lu %10u\n",
dev->name,
stats->rx_bytes,
stats->rx_packets, stats->rx_errors,
@@ -1763,7 +1765,8 @@
stats->tx_fifo_errors, stats->collisions,
stats->tx_carrier_errors + stats->tx_aborted_errors +
stats->tx_window_errors + stats->tx_heartbeat_errors,
- stats->tx_compressed);
+ stats->tx_compressed,
+ q->stats.drops);
else
size = sprintf(buffer, "%6s: No statistics available.\n",
dev->name);
@@ -1785,7 +1788,7 @@
size = sprintf(buffer,
"Inter-| Receive | Transmit\n"
- " face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed\n");
+ " face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed Qdiscdrops\n");
pos += size;
len += size;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Sep 07 2002 - 22:00:15 EST