Skip to contentSkip to conversation

ppc405_dma warnings

DISCUSSION David Gibson 2 participants 3 messages Mar 13, 2002 – Mar 14, 2002

lore
Share & tools
lore
#1 lore raw
In reply to David Gibson
Quoted 19 lines · David Gibson

Armin,

ppc405_dma.c has several constructs like:

	p_dma_ch->polarity = polarity & GET_DMA_POLARITY(0);
where GET_DMA_POLARITY() is defined:
	#define GET_DMA_POLARITY(chan) DMAReq##chan##_ActiveLow | \
		DMAAck##chan##_ActiveLow | EOT##chan##_ActiveLow

Note the macro has no parentheses around its definition, which is unusual. IIRC C precedence will mean this code turns out as:

	p_dma_ch->polarity = (polarity & DMAReq0_ActiveLow) | \
		DMAAck0_ActiveLow | EOT0_ActiveLow

which looks wrong to me and if correct is an insane use of a macro. It's also giving compiler warnings.

I'm guessing that the macro should just get parentheses, but that's a semantic change so I don't want to make it without knowing for sure. Could you check this?

Quoted 10 lines

--
David Gibson | For every complex problem there is a
[email protected] | solution which is simple, neat and
| wrong. -- H.L. Mencken
http://www.ozlabs.org/people/dgibson

Thanks.. I missed the parentheses. The macro is there to allow me to use the same driver for the 405gp dma and the stb0xxxx cpu since the the stbxxxx does not have a polarity. I could have poluted the driver with #ifdefs or mask the deltas with a few macros. I went with macros.

armin

PS I am insane

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

#0 lore raw

Armin,

ppc405_dma.c has several constructs like:

	p_dma_ch->polarity = polarity & GET_DMA_POLARITY(0);
where GET_DMA_POLARITY() is defined:
	#define GET_DMA_POLARITY(chan) DMAReq##chan##_ActiveLow | \
		DMAAck##chan##_ActiveLow | EOT##chan##_ActiveLow

Note the macro has no parentheses around its definition, which is unusual. IIRC C precedence will mean this code turns out as:

	p_dma_ch->polarity = (polarity & DMAReq0_ActiveLow) | \
		DMAAck0_ActiveLow | EOT0_ActiveLow

which looks wrong to me and if correct is an insane use of a macro. It's also giving compiler warnings.

I'm guessing that the macro should just get parentheses, but that's a semantic change so I don't want to make it without knowing for sure. Could you check this?

--
David Gibson | For every complex problem there is a
[email protected] | solution which is simple, neat and
| wrong. -- H.L. Mencken
http://www.ozlabs.org/people/dgibson

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

#2 lore raw
In reply to Armin
Quoted 16 lines · Armin
Quoted 15 lines · David Gibson

Armin,

ppc405_dma.c has several constructs like:

	p_dma_ch->polarity = polarity & GET_DMA_POLARITY(0);
where GET_DMA_POLARITY() is defined:
	#define GET_DMA_POLARITY(chan) DMAReq##chan##_ActiveLow | \
		DMAAck##chan##_ActiveLow | EOT##chan##_ActiveLow

Note the macro has no parentheses around its definition, which is unusual. IIRC C precedence will mean this code turns out as:

	p_dma_ch->polarity = (polarity & DMAReq0_ActiveLow) | \
		DMAAck0_ActiveLow | EOT0_ActiveLow

which looks wrong to me and if correct is an insane use of a macro. It's also giving compiler warnings.

Quoted 8 lines
Quoted 3 lines

I'm guessing that the macro should just get parentheses, but that's a semantic change so I don't want to make it without knowing for sure. Could you check this?

Thanks.. I missed the parentheses. The macro is there to allow me to use the same driver for the 405gp dma and the stb0xxxx cpu since the the stbxxxx does not have a polarity. I could have poluted the driver with #ifdefs or mask the deltas with a few macros. I went with macros.

Seems reasonable. Will you put in the missing parentheses, or should I?

Quoted 3 lines

armin

PS I am insane

That may be true, but accidentally missing the () doesn't prove it. Deliberately missing the () would :-)

--
David Gibson | For every complex problem there is a
[email protected] | solution which is simple, neat and
| wrong. -- H.L. Mencken
http://www.ozlabs.org/people/dgibson

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/