parport_pc: Fix subscription bugs
This patch fixes array subscription bugs in the parport_pc driver.
drivers/parport/parport_pc.c: In function ‘parport_irq_probe’:drivers/parport/parport_pc.c:1589: warning: array subscript is above array bounds drivers/parport/parport_pc.c: In function ‘parport_pc_probe_port’:drivers/parport/parport_pc.c:1579: warning: array subscript is above array bounds
The patch also fixes a few other array bugs, which the compiler was unable to find. Coding style violations are also fixed.
---
These are real bugs, not just bogus compiler warnings.
drivers/parport/parport_pc.c +18 -16
@@ -1243,23 +1243,23 @@ static void __devinit show_parconfig_smsprintk (KERN_INFO "SMSC LPT Config: enabled=%s power=%s\n" , (cr23*4 >=0x100 ) ?"yes" :"no" , (cr1 &4 ) ?"yes" :"no" );printk (KERN_INFO "SMSC LPT Config: Port mode=%s, EPP version =%s\n" , (cr1 &0x08 ) ?"Standard mode only (SPP)" : modes[cr4 &0x03 ], (cr4 &0x40 ) ?"1.7" :"1.9" ); } - +/* Heuristics ! BIOS setup for this mainboard device limits the choices to standard settings, i.e. io-address and IRQ are related, however DMA can be 1 or 3, assume DMA_A=DMA1, DMA_C=DMA3 (this is true e.g. for TYAN 1564D Tomcat IV) */ -if (cr23*4 >=0x100 ) {/* if active */ -while ((superios[i].io!=0 ) && (i<NR_SUPERIOS )) +if (cr23 *4 >=0x100 ) {/* if active */ +while ((i <NR_SUPERIOS ) && (superios[i].io !=0 )) i++; -if (i==NR_SUPERIOS ) +if (i ==NR_SUPERIOS ) {printk (KERN_INFO "Super-IO: too many chips!\n" ); -else { + }else {int d;switch (cr23*4 ) {case 0x3bc : superios[i].io =0x3bc ; superios[i].irq =7 ;break ;
@@ -1329,18 +1329,18 @@ static void __devinit show_parconfig_winprintk ("dma=%d\n" ,cr74 &0x07 );printk (KERN_INFO "Winbond LPT Config: irqtype=%s, ECP fifo threshold=%d\n" , irqtypes[crf0>>7 ], (crf0>>3 )&0x0f );printk (KERN_INFO "Winbond LPT Config: Port mode=%s\n" , modes[crf0 &0x07 ]); } -if (cr30 &0x01 ) {/* the settings can be interrogated later ... */ -while ((superios[i].io!=0 ) && (i<NR_SUPERIOS )) +if (cr30 &0x01 ) {/* the settings can be interrogated later ... */ +while ((i <NR_SUPERIOS ) && (superios[i].io !=0 )) i++; -if (i==NR_SUPERIOS ) +if (i ==NR_SUPERIOS ) {printk (KERN_INFO "Super-IO: too many chips!\n" ); -else { + }else { superios[i].io = (cr60<<8 )|cr61; superios[i].irq = cr70&0x0f ; superios[i].dma = (((cr74 &0x07 ) >3 ) ?PARPORT_DMA_NONE : (cr74 &0x07 )); } }
@@ -1572,30 +1572,32 @@ static void __devinit detect_and_report_ release_region(0x2e, 1); }#endif /* CONFIG_PARPORT_PC_SUPERIO */ static int get_superio_dma (struct parport *p) { -int i=0 ; -while ( (superios[i].io != p->base) && (i<NR_SUPERIOS )) +int i =0 ; + +while ((i <NR_SUPERIOS ) && (superios[i].io != p->base)) i++; -if (i!=NR_SUPERIOS ) +if (i !=NR_SUPERIOS )return superios[i].dma;return PARPORT_DMA_NONE ; }static int get_superio_irq (struct parport *p) { -int i=0 ; -while ( (superios[i].io != p->base) && (i<NR_SUPERIOS )) +int i =0 ; + +while ((i <NR_SUPERIOS ) && (superios[i].io != p->base)) i++; -if (i!=NR_SUPERIOS ) +if (i !=NR_SUPERIOS )return superios[i].irq;return PARPORT_IRQ_NONE ; } - +/* --- Mode detection ------------------------------------- */ /* * Checks for port existence, all ports support SPP MODE * Returns:
Signature · 1 line
Greetings, Michael.
Quoted 9 lines · On Mon, 1 Jun 2009 11:00:33 +0200 Michael Buesch <[email protected]> wrote:
This patch fixes array subscription bugs in the
parport_pcdriver.
drivers/parport/parport_pc.c: In function ‘parport_irq_probe’:drivers/parport/parport_pc.c:1589: warning: array subscript is above array boundsdrivers/parport/parport_pc.c: In function ‘parport_pc_probe_port’:drivers/parport/parport_pc.c:1579: warning: array subscript is above array boundsThe patch also fixes a few other array bugs, which the compiler was unable to find. Coding style violations are also fixed.
Which makes it very hard to see what you are actually changing and whether it is a bug. It certainly looks like one but the changes muddled in with a random subset of format fixes (even in the area patched only some style is fixed) looks pretty clunky - surely a for() loop would be far cleaner.
But yes agreed - its a bug, it wants fixing and style can be sorted later.
Quoted 10 lines · On Monday 01 June 2009 11:14:40 Alan Cox wrote:
Quoted 7 lines · On Mon, 1 Jun 2009 11:00:33 +0200 Michael Buesch <[email protected]> wrote:
This patch fixes array subscription bugs in the
parport_pcdriver.
drivers/parport/parport_pc.c: In function ‘parport_irq_probe’:drivers/parport/parport_pc.c:1589: warning: array subscript is above array boundsdrivers/parport/parport_pc.c: In function ‘parport_pc_probe_port’:drivers/parport/parport_pc.c:1579: warning: array subscript is above array bounds
Quoted 7 lines
Quoted 2 lines
The patch also fixes a few other array bugs, which the compiler was unable to find. Coding style violations are also fixed.
Which makes it very hard to see what you are actually changing and whether it is a bug. It certainly looks like one but the changes muddled in with a random subset of format fixes (even in the area patched only some style is fixed)
Yeah well. If I did not fix the style violations, somebody else would have complained already that when somebody touches some code he should also fix the style issues. ;) Happened to be several times, so I chose to fix them here...
Signature · 1 line
Greetings, Michael.
Quoted 3 lines
Yeah well. If I did not fix the style violations, somebody else would have complained already that when somebody touches some code he should also fix the style issues. ;) Happened to be several times, so I chose to fix them here...
Yes Andrew likes code to become horrible and inconsistent it seems ;)
I'll post a follow up to your patch to whack the whole thing into style and switch those loops.
Quoted 8 lines · On Monday 01 June 2009 11:58:19 Alan Cox wrote:
Quoted 3 lines
Yeah well. If I did not fix the style violations, somebody else would have complained already that when somebody touches some code he should also fix the style issues. ;) Happened to be several times, so I chose to fix them here...
Yes Andrew likes code to become horrible and inconsistent it seems ;)
I'll post a follow up to your patch to whack the whole thing into style and switch those loops.
Cool. Thanks a lot Alan. :)
Signature · 1 line
Greetings, Michael.